Category: 3. Business

Reassessing a hypermetabolic splenic lesion in breast cancer: PET/CT findings and insights from multimodal imaging and multidisciplinary evaluation | Egyptian Journal of Radiology and Nuclear Medicine

In this case report, we highlight the dual-edged utility of [¹⁸F]FDG PET/CT in the management of breast cancer, illustrating both its strengths in staging and response assessment and its pitfalls related to specificity. The role of [¹⁸F]FDG PET/CT in breast cancer has been increasingly recognized, particularly for phenotypes characterized by high glycolytic activity—namely triple-negative and “HER2-enriched” subtypes [1, 5, 6]. The Phergain study prospectively demonstrated that changes in [¹⁸F]FDG uptake during neoadjuvant therapy can guide adaptive treatment strategies, improving pathological complete response rates in HER2-positive disease [7]. This study underscored the prognostic importance of metabolic response on interim PET and validated [¹⁸F]FDG PET/CT as a decision-making tool during treatment.

However, [¹⁸F]FDG PET/CT is not without limitations. Focal FDG-avid splenic lesions encompass a wide differential, including benign vascular entities such as cavernous or capillary hemangiomas, sclerosing angiomatoid nodular transformation (SANT), and less common vascular neoplasms like hemangioendothelioma [8, 9]. Other considerations include inflammatory or granulomatous disorders (e.g., sarcoidosis, tuberculosis, other mycobacterial or fungal infections), infectious abscess, hematologic involvement by lymphoma or leukemia, and metastatic disease. Several benign entities (notably SANT and atypical hemangiomas) can show variable FDG uptake and overlapping cross-sectional features, which complicate noninvasive diagnosis [10, 11]. Histopathology remains the reference standard for definitive diagnosis, but percutaneous splenic biopsy carries bleeding and sampling-error risks and may be impractical for small lesions—therefore, tailored integration of imaging features, clinical context, and multidisciplinary review is often required when tissue is not available.

A major drawback of [¹⁸F]FDG PET/CT is its limited specificity. [¹⁸F]FDG uptake is not tumor‐specific, and many benign processes—including inflammation, infection, and vascular lesions—can exhibit significant tracer accumulation. Large retrospective series and case reviews have similarly emphasized that splenic involvement by breast cancer is an uncommon event, with a pooled prevalence well below 1% across thousands of PET/CT studies [12, 13]. Consequently, the evidence of an FDG-avid splenic nodule in a breast cancer patient almost always prompts consideration of benign etiologies, with vascular anomalies such as hemangiomas ranking high on the differential.

In this context, MRI and contrast CT provide complementary morphologic and enhancement information to PET metabolic data. Typical cavernous hemangiomas usually appear markedly T2-hyperintense and demonstrate peripheral nodular enhancement with progressive centripetal fill-in on delayed phases (CT or MRI), while they may be variably FDG-avid [14]. SANT, although rare, often shows a characteristic spoke-wheel or star-shaped enhancement with a central fibrous scar and persistent delayed enhancement. In addition, on MRI, SANT may be heterogeneous with lower T2 signal centrally [15]. On PET/CT, both SANT and atypical vascular lesions can show increased [¹⁸F]FDG uptake, limiting specificity [10, 11]. Thus, enhancement pattern, diffusion behavior, and clinical context are the most useful MRI/CT discriminators, while PET adds sensitivity for metabolic activity but not reliable lesion-type specificity. In this regard, quantitative PET-derived parameters may help in distinguishing benign from malignant lesions. In particular, an SUV threshold of 2.3 has been reported to differentiate benign from malignant lesions, although this finding has not been validated in larger cohorts [8]. In our patient, in fact, the increased and focal tracer uptake (SUVmax 6.5) in the splenic lesion was misleading, as it suggested a neoplastic rather than a benign origin.

Our patient’s baseline staging PET/CT revealed an isolated hypermetabolic splenic lesion (SUVmax 6.5) with corresponding CT findings of a 16 mm hypodense nodule possessing a central low-density core and peripheral rim enhancement—features that are often worrisome for metastasis. Given the uncommon nature of splenic metastases in breast cancer, we pursued multimodal imaging. MRI characterization demonstrated no diffusion restriction and progressive contrast filling on delayed sequences, radiologically favoring an atypical hemangioma. Indeed, focal [¹⁸F]FDG uptake in atypical hemangiomas has been documented in both vertebral and hepatic locations, further complicating interpretation [16, 17]. In these reports, benign vascular tumors displayed intense [¹⁸F]FDG avidity akin to malignant lesions, likely reflecting the high endothelial cell turnover or inflammatory milieu within the lesion. What made our case particularly challenging was the combination of intense [¹⁸F]FDG uptake with CT features mimicking metastatic disease—a presentation that even experienced observers found disconcerting.

The therapeutic course provided an opportunity to further clarify the lesion’s nature. Following the first six cycles of neoadjuvant carboplatin, docetaxel, pertuzumab, and trastuzumab, repeat PET/CT demonstrated complete metabolic resolution of the splenic focus alongside reduction of the nodule to 5 mm on CT and preserved benign imaging characteristics on MRI. Such volume reduction of benign splenic lesions post‐therapy has been sporadically noted in the literature: one case series reported shrinkage of a splenic hemangioma following systemic chemotherapy for lymphoma [18]. However, to our knowledge, this is the first documented case of an atypical hemangioma in the spleen of a breast cancer patient evaluated in a true multimodal fashion—PET, CT, and MRI—both before and after neoadjuvant chemotherapy, with a demonstrated complete metabolic response.

The observed treatment-associated involution of the splenic lesion could be attributed to non‐specific effects of cytotoxic and targeted therapies on the vascular endothelium, inflammatory stroma, or associated macrophages within the hemangioma [19]. Such phenomena underscore the necessity of caution when interpreting post-therapy reductions in size or metabolic activity, as benign lesions may mimic true tumor response. In our patient, multidisciplinary review—bringing together nuclear medicine physicians, radiologists, oncologists, and surgeons—was pivotal in integrating imaging findings with clinical context, thereby avoiding overtreatment based on a false-positive PET result.

Despite these insights, several limitations must be acknowledged. First, the diagnosis of atypical hemangioma remains presumptive, as histologic confirmation was not available. Although percutaneous biopsy of small splenic lesions carries risks of hemorrhage and sampling error, tissue diagnosis remains the gold standard. Second, while [¹⁸F]FDG PET/CT remains widely accessible, its lack of specificity in differentiating benign from malignant vascular splenic lesions suggests a potential role for more selective radiotracers. Novel agents such as fibroblast activation protein inhibitor (FAPI) analogues have demonstrated low-background splenic uptake and high tumor-to-background ratios in breast cancer, offering promise for improved specificity [20, 21]. Preliminary studies of HER2-targeted agents labeled with positron emitters (e.g., ⁸⁹Zr-trastuzumab) have also shown utility in assessing HER2 expression in metastatic lesions while sparing benign tissues [22]. These targeted tracers could potentially reduce false-positive findings in the spleen by exploiting receptor-based uptake rather than glycolytic activity alone.

Moreover, advanced MRI techniques—such as dynamic contrast-enhanced perfusion imaging, diffusion tensor imaging, and MR elastography—may provide additional tissue characterization metrics to distinguish benign vascular anomalies from metastatic deposits. Radiomics and machine learning approaches using multiparametric MRI and PET features may further refine diagnostic accuracy [23].

Continue Reading

October 25, 2025
Degenerative rotator cuff tears in correlation with different anatomic shoulder parameters on MRI | Egyptian Journal of Radiology and Nuclear Medicine

The study included 67 participants with an average age of 41.58 years (SD = 16.77). The age of participants ranged from 18 to 69 years, with a median age of 44 years. The interquartile range (IQR) revealed that 25% of the participants were aged 23 years or younger, while 75% were aged 57 years or younger, indicating a wide distribution of age among the participants. The gender distribution revealed that 59.7% of the patients were male (n = 40), while 40.3% were female (n = 27). Regarding the side affected, the majority of the patients had the right side affected, accounting for 53.7% (n = 36), whereas 46.3% (n = 31) had the left side affected. When categorized based on the type of tear, 49.3% of the patients were in the control group (n = 33), 26.9% had a partial tear (n = 18), and 23.9% had a full tear (n = 16) Table 1.

Table 1 Demographical characteristics of the study patients (N = 67)

The analysis revealed a strong positive correlation between the Critical Shoulder Angle (CSA) and the presence of RCT (r = 0.827, P < .0001), indicating that higher CSA values were associated with a higher likelihood of RCT. Similarly, the Acromial Index (AI) also showed a significant positive correlation (r = 0.695, P < .0001). Conversely, the Lateral Acromial Angle (LAA) and Acromiohumeral Distance (AHD) were negatively correlated with the presence of RCT (r = − 0.542, P < .0001; r = − 0.413, P = .001) respectively, suggesting that lower values of these parameters were associated with an increased risk of RCT Table 2.

Table 2 Evaluation of the relationship between shoulder parameters and the presence of rotator cuff tear (RCT)

The comparative analysis of shoulder parameters between the Control and Rotator Cuff Tear (RCT) group. For the Critical Shoulder Angle (CSA), the Control group demonstrated a mean value of 31.52 ± 2.14, with a median of 32.00 and a range of 28 to 35. In contrast, the RCT group exhibited a significantly higher mean CSA of 37.12 ± 2.27, with a median of 37.00 and a range of 30 to 43. The Mann–Whitney U test revealed a highly significant difference between the groups (P < 0.001, Z = − 6.717).

Regarding the Lateral Acromial Angle (LAA), the Control group had a mean of 82.82 ± 3.47 and a median of 83.00, with values ranging from 78 to 91. The RCT group, however, had a lower mean LAA of 77.65 ± 4.35, with a median of 78.00 and a range of 70 to 86. This difference was also statistically significant (P < 0.001, Z = − 4.405).

The Acromial Index (AI) was higher in the RCT group, with a mean of 0.702 ± 0.053 and a median of 0.690, compared to the Control group’s mean of 0.617 ± 0.046 and a median of 0.630. The range for AI in the Control group was 0.52 to 0.68, while in the RCT group, it ranged from 0.60 to 0.80. This difference was statistically significant, with a p-value of <0.001 (Z = − 5.644).

Lastly, the Acromiohumeral Distance (AHD) was found to be significantly lower in the RCT group, with a mean of 6.22 ± 1.54 and a median of 6.40, compared to the Control group’s mean of 7.45 ± 1.09 and a median of 7.70. The AHD ranged from 5.00 to 8.90 in the Control group and from 2.50 to 8.60 in the RCT group. The difference between the groups was statistically significant (P = 0.001, Z = − 3.357).

These findings indicate significant variations in shoulder parameters between individuals with rotator cuff tears and those without, suggesting that rotator cuff tears were associated with distinct alterations in anatomical measurements Table 3.

Table 3 Comparison of shoulder parameters between control and rotator cuff tear group

The effectiveness of various shoulder parameters in determining the presence of rotator cuff tears (RCT) was assessed using the Area Under the Curve (AUC) from receiver operating characteristic (ROC) analysis. The Critical Shoulder Angle (CSA) exhibited the highest AUC of 0.975, with a 95% confidence interval (CI) ranging from 0.933 to 1.000, and a standard error (SE) of 0.021. This AUC indicated an exceptional ability of CSA to discriminate between individuals with and without RCT, with the p-value of .000 confirming statistical significance.

Critical shoulder angle (CSA) at 34 degrees: Sensitivity: 0.971 (97.1%): At a cut-off value of 34 degrees, the CSA correctly identified 97.1% of individuals with the condition (true positives). This high sensitivity implied that CSA was highly effective in detecting the condition, minimizing the number of false negatives. Specificity: 0.909 (90.9%): The specificity of 90.9% indicated that the CSA correctly identified 90.9% of individuals without the condition (true negatives). This demonstrated that the CSA had a good ability to correctly exclude those who did not have the condition, reducing false positives.

The Acromial Index (AI) also demonstrated strong discriminative power with an AUC of 0.900 (95% CI 0.829-0.971) and a SE of 0.036. The significance of this parameter is supported by a p-value of 0.000, suggesting it was a highly effective metric for detecting RCT. Acromial Index (AI) at 0.66: Sensitivity: 0.765 (76.5%): At a cut-off of 0.66, AI correctly identified 76.5% of individuals with the condition Specificity: 0.879 (87.9%): The specificity of 87.9% showed that AI effectively identified 87.9% of those without the condition.

In contrast, the Lateral Acromial Angle (LAA) had a markedly lower AUC of 0.188 (95% CI 0.087-0.289) with a SE of 0.052. Despite the statistically significant p-value of .000, this low AUC reflected a poor ability of LAA to differentiate between RCT and control groups, indicating limited utility in clinical settings.

Similarly, the Acromiohumeral Distance (AHD) showed an AUC of 0.262 (95% CI 0.141-0.382) with a SE of 0.061, and a p-value of .001. Although statistically significant, the low AUC suggested that AHD was not a strong predictor of RCT, demonstrating limited effectiveness compared to CSA and AI. Overall, CSA and AI were identified as the most effective parameters for determining the presence of rotator cuff tears, while LAA and AHD were less effective Table 4.

Table 4 Effectiveness of shoulder parameters to determine the presence of rotator cuff tear

CSA (Critical shoulder angle): The mean CSA for the control group was 31.52 ± 2.14 degrees, with a median of 32.00 (range 28–35). The partial tear group showed a significantly higher mean CSA of 36.33 ± 2.25 degrees and a median of 36.00 (range 30–41). The full tear group had the highest mean CSA at 38.00 ± 2.00 degrees, with a median of 37.50 (range 36–43). The Kruskal–Wallis test revealed a statistically significant difference between groups (P = 0.000, KW = 47.210), indicating that CSA values increased with the severity of rotator cuff tear.

LAA (Lateral acromial angle): The mean LAA for the control group was 82.82 ± 3.47 degrees, with a median of 83.00 (range 78–91). The partial tear group had a lower mean LAA of 77.44 ± 4.23 degrees and a median of 78.00 (range 70–86). The full tear group showed a similar mean LAA of 77.88 ± 4.60 degrees, with a median of 78.50 (range 70–86). The Kruskal–Wallis test indicated a significant difference among groups (P = 0.000, KW = 19.584), suggesting a reduction in LAA with the severity of rotator cuff tear.

AI (Acromial index): The control group had a mean AI of 0.62 ± 0.05, with a median of 0.63 (range 0.52–0.68). The partial tear group showed a higher mean AI of 0.70 ± 0.05 and a median of 0.70 (range 0.61–0.80). The full tear group exhibited the highest mean AI of 0.71 ± 0.06, with a median of 0.69 (range 0.60–0.80). The Kruskal–Wallis test results showed a significant difference (P = 0.000, KW = 31.951), indicating an increase in AI with the severity of the tear.

AHD (Acromiohumeral distance): For the control group, the mean AHD is 7.45 ± 1.09 mm, with a median of 7.70 (range 5.00–8.90). The partial tear group showed a slightly lower mean AHD of 7.19 ± 0.86 mm and a median of 7.15 (range 5.50–8.60). The full tear group had the lowest mean AHD at 5.13 ± 1.41 mm, with a median of 5.10 (range 2.50–8.40). The Kruskal–Wallis test showed a significant difference (P = 0.000, KW = 23.675), suggested a decrease in AHD with increasing severity of the rotator cuff tear. In summary, the analysis using the Kruskal–Wallis test showed significant differences in all four shoulder parameters (CSA, LAA, AI, and AHD) among the different tear size subgroups. Specifically, the CSA and AI values tended to increase with the severity of the tear, while the AHD and LAA decreased. These findings highlighted the impact of tear severity on shoulder parameters Table 5.

Table 5 Comparison of shoulder parameters among tear size subgroups

CSA (Critical shoulder angle): Control vs Partial tear: The AUC was 0.953 with a 95% confidence interval of 0.876 to 1.000, and a p-value of 0.000. This indicated excellent discriminative ability for CSA in distinguishing between control and partial tear groups. Control versus Full Tear: The AUC was 1.000 with a 95% confidence interval of 1.000 to 1.000, and a p-value of 0.000. This perfect AUC suggests CSA is highly effective in distinguishing between control and full tear groups. Partial vs Full Tear: The AUC is 0.738 with a 95% confidence interval of 0.570 to 0.906, and a p-value of 0.000. This indicates good discriminative ability for CSA in differentiating between partial and full tear groups.

AI (Acromial index): Control vs Partial tear: The AUC was 0.890 with a 95% confidence interval of 0.796 to 0.983, and a p-value of 0.000. This suggested that AI was effective in differentiating between control and partial tear groups. Control vs Full Tear: The AUC was 0.912 with a 95% confidence interval of 0.822 to 1.000, and a p-value of 0.000. This indicated that AI was highly effective in distinguishing between control and full tear groups. Partial vs Full Tear: The AUC was 0.540 with a 95% confidence interval of 0.342 to 0.738, and a p-value of 0.692. This AUC suggested poor discriminative ability of AI in differentiating between partial and full tear groups.

LAA (Lateral acromial angle): Control vs Partial tear: The AUC was 0.166 with a 95% confidence interval of 0.044 to 0.287, and a p-value of 0.000. This very low AUC indicated poor performance of LAA in distinguishing between control and partial tear groups. Control vs Full Tear: The AUC was 0.213 with a 95% confidence interval of 0.072 to 0.354, and a p-value of 0.001. This low AUC suggested that LAA was not effective in distinguishing between control and full tear groups. Partial vs Full Tear: The AUC was 0.538 with a 95% confidence interval of 0.340 to 0.737, and a p-value of 0.704. This AUC suggested poor discriminative ability of LAA in differentiating between partial and full tear groups.

AHD (Acromiohumeral distance): Control vs Partial tear: The AUC was 0.396 with a 95% confidence interval of 0.234 to 0.559, and a p-value of 0.225. This indicated poor discriminative ability of AHD in distinguishing between control and partial tear groups. Control vs Full Tear: The AUC was 0.110 with a 95% confidence interval of 0.000 to 0.221, and a p-value of 0.000. This very low AUC suggested that AHD was not effective in differentiating between control and full tear groups. Partial vs Full Tear: The AUC was 0.085 with a 95% confidence interval of 0.000 to 0.198, and a p-value of 0.000. This very low AUC indicated poor performance of AHD in distinguishing between partial and full tear groups Table 6.

Table 6 Effectiveness of shoulder parameters in predicting tear size

Continue Reading

October 25, 2025
UK's Petrofac lines up administrator, Sky News reports – Reuters
1. UK’s Petrofac lines up administrator, Sky News reports Reuters
2. TenneT’s 2 GW Contract Termination Derails Petrofac’s Restructuring Plans offshoreWIND.biz
3. Administrators lined up for North Sea oilfield services group Petrofac Sky News
4. Ministers on alert as North Sea supplier scrambles to avert collapse The Telegraph
5. Petrobras and Equinor secure pre-salt star in competitive Brazil bid round Upstream Online
Continue Reading
October 25, 2025
LAUNCHED: ASEAN Investment Report 2025: FDI as a Key Catalyst for Advancing Regional Supply Chains

ASEAN shall develop friendly relations and mutually beneficial dialogues, cooperation and partnerships with countries and sub-regional, regional and international organisations and institutions. This includes external partners, ASEAN entities, human rights bodies, non-ASEAN Member States Ambassadors to ASEAN, ASEAN committees in third countries and international organisations, as well as international / regional organisations.

Continue Reading

October 25, 2025
Banks Rocked by ‘Extreme’ Car Loan Costs Gear Up for FCA Fight

The UK’s biggest banks are gearing up for yet another fight with regulators over how they’ll compensate consumers who were missold car loans — even after they set aside an additional £1.5 billion to resolve the saga in recent weeks.

Barclays Plc on Wednesday said it had roughly quadrupled the amount of cash it has set aside to compensate customers who were impacted by the scandal. One day later, Lloyds Banking Group Plc saw its pre-tax profit in the third quarter slump 36% because of an additional £800 million charge tied to the matter.

Continue Reading

October 25, 2025
HAYA Therapeutics: Pioneering Cell Therapy Innovations
Samir Ounzain, CEO of HAYA Therapeutics, is a molecular biologist with over 15 years of experience exploring the dark matter of the genome and its roles in development and disease.

Thomas Kern / SWI swissinfo
Generated with artificial intelligence.
Samir Ounzain, CEO of science start-up HAYA Therapeutics, talks about the limits of entrepreneurship in Switzerland, entering the US market and how his therapy for heart disease could change the world of medicine.

This content was published on

October 25, 2025 – 10:00
As a photo editor I am responsible for the editorial use of photography at SWI swissinfo.ch and our collaborations with photographers. When the opportunity arises, I take a camera and accompany one of our journalists.
I trained as a photographer in Zürich and began working as a photojournalist in 1989. I was a founder of the Swiss photographers’ agency Lookat Photos in 1990. A two-time World Press Award winner, I have also been awarded several Swiss national scholarships. My work has been widely exhibited and it is represented in various collections.

More from this author

Multimedia
Founded in 2019, HAYA Therapeutics aims to pioneer an approach to treating disease by turning sick cells back into healthy ones. The start-up is currently developing a very specific therapy for heart failure and is about to begin the three challenging phases of clinical trials. If successful, the novel approach could be used to treat a wide range of common, chronic, and age-related diseases.

HAYA Therapeutics has so far raised around $90 million (CHF72 million) and benefitted from investor confidence, but the next steps to success could be challenging, including gaining market authorisation, CEO and co-founder Samir Ounzain tells Swissinfo from his headquarters in Lausanne.

Thomas Kern / SWI swissinfo

Swissinfo: You are British, you studied in the United Kingdom but you decided to set your start-up in Switzerland. How come?

Samir Ounzain: Switzerland is an excellent place to translate academic ideas into commercial products. As a start-up, we received a great deal of support. We are also located at the BiopôleExternal link campus, near Lausanne, which is fully dedicated to life sciences. This campus brings together ambitious start-ups, major multinational companies, and research institutions, while providing state-of-the-art research facilities.

We also benefit from the long legacy of Switzerland’s pharmaceutical industry. In my view, the main asset here is access to talent – both those trained in Switzerland and Europeans attracted by the country’s high quality of life and professional opportunities.

Swissinfo: You also have a structure in the United States focused largely on fund raising and then market access. Why was this important to you?

S.O.: Our goal is to make the fastest and most significant impact on patients in need by pioneering a completely new approach to drug discovery and development. For this reason, we have taken a global perspective: we seek the very best in the world in terms of partners, suppliers, talent, financing opportunities, and market attractiveness – rather than focusing only on Switzerland.

That said, we are very satisfied with our Swiss-American structure. As mentioned, Switzerland remains an excellent place to attract talent and is highly respected in our field. The United States, meanwhile, offers unmatched opportunities in terms of financing, seasoned biotech operators, and market size.

Switzerland and Europe still have room for improvement when it comes to scaling up their start-ups. It is very rare for companies that are less than ten years old to reach ‘unicorn’ status (valued at over $1 billion), particularly in the life sciences sector.

Swissinfo: Swiss tax laws require company founding owners to pay heavy taxes based on the virtual valuation of their start-ups. Is this a hurdle for your future expansion in Switzerland?

S.O.: Yes, Switzerland is one of the few countries worldwide that taxes wealth. For start-up founding owners, this tax can amount to around 1% of the valuation set by external investors. For instance, if a start-up is valued at CHF1 billion, its shareholders – including the founders, who typically receive modest salaries – must pay every year CHF10 million in wealth tax.

Although the tax shield scheme [which caps the cantonal taxes at 60% of taxable income], mitigates the impact of this issue, remaining in Switzerland can still be challenging for entrepreneurs. The problem lies in the sharp discrepancy between a start-up’s virtual valuation, which is based on its potential for long-term success, and the actual liquidity available to its founders. Finding a solution to this enormous challenge would benefit the Swiss start-up ecosystem.

Thomas Kern / SWI swissinfo

Swissinfo: You raised about $90 million. Like other promising start-ups, your investors are foreign, namely from the US and EU countries. Does this encourage you to relocate closer to them?

S.O.: In Switzerland, early-stage financial support, such as seed financing, is widely available. However, later-stage growth capital is much scarcer. In our case, for example, most of our current investors are based abroad. In our most recent financing round, which raised $65 million, the lead investors were Sofinnova Partners (with offices in Paris, Milan, and London) and Earlybird Venture Capital (with offices in Berlin, London, Milan, and Munich).

Generally, international investors view Switzerland positively in terms of reliability and innovation and do not systematically push companies to relocate. However, at HAYA Therapeutics, we decided to establish an affiliate in San Diego to be closer to the US market, experienced biotech operators, and US investors. This last point is crucial: American investors typically have a significant appetite for risk and bold ideas.

Swissinfo: Bringing a new medicine to market usually costs around $1 billion, largely due to very expensive and extensive phase 3 clinical trials. How do you plan to finance this?

S.O.: Indeed, significant funding will be required in the future. We plan to begin the first phase of clinical trials for our lead product early next year, with the immediate goal of demonstrating safety and early signs of efficacy. If successful, several financing options should open up, including growth funds, partnerships with large pharmaceutical companies, or an initial public offering (IPO) – likely on Nasdaq in New York. Our awards, particularly our designation as a Technology Pioneer by the World Economic Forum (WEF), enhance our visibility and credibility, thereby strengthening our position in securing the necessary financing.

HAYA Therapeutics: Key Figures

Kai Reusser / SWI swissinfo.ch

Swissinfo: It is prohibitively expensive for a start-up to seek worldwide regulatory approval. Where do you plan to seek market authorisation for your therapy first?

S.O.: Each country or bloc (for example the European Union) has its own regulatory framework. Targeting all of them at once is indeed too costly for a start-up. Our current priority is to conduct clinical trials aligned with US requirements of the Food and Drug Administration (FDA), since the US is the largest market for us.

We aim to be capital-efficient. Therefore, we are considering conducting parts of our clinical trials in jurisdictions that are more cost-effective and offer faster patient recruitment. This would allow us to accelerate our development timelines while remaining compliant with regulatory requirements.

Thomas Kern / SWI swissinfo

Swissinfo: Your lead product, a therapy for heart failure, targets a very specific market. Is this to maximise your chances of success?

S.O.: Our lead product, HTX-001, is a targeted therapy for heart failure, focused on non-obstructive hypertrophic cardiomyopathy. This therapy represents a novel approach: treating cells that behave abnormally. Importantly, if we can prove that modifying cellular states delivers positive outcomes for this indication, we can then apply our methodology to a wide range of common, chronic, and age-related diseases such as hypertension, metabolic disorders, cardiovascular diseases, Alzheimer’s, and cancer.

Swissinfo: Drugs are getting more expensive to develop and bring to market. How can you ensure patients access your therapies?

S.O.: We are guided by a clear principle: bringing to market medicines that are safe, needed, effective, and accessible. Our ultimate mission is to address the unmet needs of up to ten million patients. This means our medicines must be made available at affordable prices, partly through reimbursement schemes.

As mentioned, we aim to transform the way the industry approaches drug discovery and development. The therapies we develop (based on RNA) are naturally programmable and relatively easy to scale up. Manufacturing and overall costs remain comparatively low, which reinforces our confidence that our medicines will be affordable.

Edited by Virginie Mangin/ts
Continue Reading
October 25, 2025
Got $5,000? 2 Tech Stocks to Buy and Hold for the Long Term

Microsoft’s subscription-based model continues to drive impressive growth for investors.

Netflix is a highly profitable entertainment company with great long-term prospects.

10 stocks we like better than Microsoft ›

The tech-heavy Nasdaq Composite (NASDAQINDEX: ^IXIC) has significantly outperformed the other major indices over the last decade. That streak has continued in 2025, with the Nasdaq up 19% year to date, beating the S&P 500 (SNPINDEX: ^GSPC) and Dow Jones Industrial Average (DJINDICES: ^DJI) returns of 14% and 9%, respectively.

The tech sector is full of innovative, fast-growing companies that can help you crush the market’s average return. If you are fortunate to have extra cash you don’t need for near-term living expenses, here are two tech stocks that can multiply your savings over the long term.

Image source: Getty Images.

Microsoft (NASDAQ: MSFT) is about as rock-solid as they come. It powers services that people use every day, from Windows and Office to Xbox gaming. But it’s also impacting the future of computing with its fast-growing enterprise cloud service, Microsoft Azure.

The stock has delivered market-beating returns over the last decade, as Microsoft shifted from its PC dependency to a subscription-based model. That strategic shift not only boosted its revenue growth, but its profitability, too.

Microsoft posted impressive revenue growth of 18% year over year in the company’s June-ending fiscal fourth quarter. Analysts expect the company to maintain 14% annual growth over the next few years. For fiscal 2025, revenue from cloud grew 23% year over year, reaching $168 billion over the last year. This reflects tremendous demand for Microsoft’s software and enterprise cloud services.

The growth in these services has swelled Microsoft’s bottom line. Its profit margin is stellar at 36%. The company produced $71 billion in free cash flow on $281 billion of total revenue over the last year.

This soaring profitability will continue to fund investments in artificial intelligence (AI). Its Copilot assistant has gained wide adoption, with 100 million monthly users across consumer and enterprise.

Microsoft is also investing in quantum computing, which promises to be the next leg of growth beyond AI. This is one of the most dominant tech companies in the world, and it’s not short of growth opportunities.

All this means Microsoft is a quality growth stock to park a few thousand dollars for the long term.

Image source: Netflix.

Netflix (NASDAQ: NFLX) is a highly profitable entertainment powerhouse. Its new releases can hit impressive viewership numbers that drive media buzz. But what ultimately makes Netflix a great investment is stellar financials. Netflix turns recurring revenues from subscriptions into Microsoft-like margins.

Recent releases have garnered impressive viewership numbers. The recent launch of Happy Gilmore 2 broke viewership records with 2.9 billion minutes watched during opening weekend. It also had a massive hit with K-Pop Demon Hunters reaching 266 million views within 11 weeks.

Despite generating $41 billion in trailing-12-month revenue with more than 300 million paid households, Netflix continues to grow the top line at double-digit rates. Company guidance calls for revenue to be up 17% year over year in Q3.

Its margins don’t resemble a typical entertainment business. The company generated $10 billion in net profit over the last year, representing a margin of nearly 25%. Growing profits is a key driver of shareholder returns over time, and management is continuing to invest in ways that will make the company more profitable and valuable down the road.

One example of where it’s investing is advertising. This relatively new revenue stream is set to double this year, which reflects strong demand for the more affordable ad-supported plans. Netflix is offering incredible value that should sustain strong growth. There are tremendous opportunities for Netflix to drive growth across international markets and stretch its addressable market with interactive content and documentaries.

Despite the stock’s monster run, analysts are projecting 24% annualized earnings growth in the coming years. This should lead to excellent returns for investors.

Before you buy stock in Microsoft, consider this:

The Motley Fool Stock Advisor analyst team just identified what they believe are the 10 best stocks for investors to buy now… and Microsoft wasn’t one of them. The 10 stocks that made the cut could produce monster returns in the coming years.

Consider when Netflix made this list on December 17, 2004… if you invested $1,000 at the time of our recommendation, you’d have $600,550!* Or when Nvidia made this list on April 15, 2005… if you invested $1,000 at the time of our recommendation, you’d have $1,116,616!*

Now, it’s worth noting Stock Advisor’s total average return is 1,032% — a market-crushing outperformance compared to 192% for the S&P 500. Don’t miss the latest top 10 list, available with Stock Advisor, and join an investing community built by individual investors for individual investors.

See the 10 stocks »

*Stock Advisor returns as of October 20, 2025

John Ballard has no position in any of the stocks mentioned. The Motley Fool has positions in and recommends Microsoft and Netflix. The Motley Fool recommends the following options: long January 2026 $395 calls on Microsoft and short January 2026 $405 calls on Microsoft. The Motley Fool has a disclosure policy.

Got $5,000? 2 Tech Stocks to Buy and Hold for the Long Term was originally published by The Motley Fool

Continue Reading

October 25, 2025

Working Memory Load–Dependent Cortical Mechanism of Distraction Anal

Introduction

As an important cognitive factor, attention plays an important role in pain processing.^1–5 Distraction through synchronized activities may reduce pain sensitivity. According to the theory of limited cognitive resources in psychology,⁶ the two-way interaction between pain and cognition can be explained from “bottom-up” and “top-down” mechanisms.⁷ Acute pain triggers bottom-up automated attentional capture through thalamo-insular pathways that prioritize cognitive resources to initiate protective responses. Experimental evidence has shown that nociceptive input can divert attention from the current task to the nociceptive stimuli. Irrelevant nociceptive stimuli interfere with cognitive performance by competing for limited attentional resources.⁸ Simultaneously, top-down attention control can modulate pain perception through distraction. When attention is allocated to the tasks without pain, the processing of nociceptive signals is suppressed due to resource competition.^9–11

Notably, working memory (WM) constitutes an essential component of executive function, optimizing attention by maintaining memory traces of attention sets and shielding goal-directed processes from interference during task execution.¹¹ It is essential in modulating the attention–pain interaction, primarily by balancing cognitive resources between nociceptive distraction and goal-directed attention.^12–16 The prefrontal cortex (PFC), involved in both executive functioning and pain processing, may experience competition for limited neural resources in pain distraction.¹⁷ Effective cognitive control of pain requires diverting attention from nociceptive stimuli and maintaining task focus through WM engagement. Different WM load may directly modulate efficiency in attention regulation of pain through resource allocation mechanism.¹⁸ Recent studies have shown that high cognitive load increases demand on attention, reducing available resources for processing extraneous stimuli and preserving task performance.^19,20 Although Deldar Z et al²⁰ have demonstrated that performing high load WM tasks may increase the allocation of attention resources and reduce pain perception, they also have reported that high load tasks may diminish the contribution of WM to distraction analgesia due to factors such as cognitive effort and ceiling effect, ultimately reducing the distraction analgesic effect. The competitive occupation of attention resources by cognitive fatigue can lead to increased distraction and a reduced ability to alleviate pain.²¹ When WM capacity reaches the limit under high-load conditions, it results in a ceiling effect on WM-based pain inhibition.²⁰ These factors collectively imply a nonlinear relationship between WM load and distraction analgesia. The effect of WM load on distraction analgesia and its underlying mechanism remains further investigation.

Neurophysiological studies have shown that distraction analgesia involves decreased neural activity in pain-processing regions such as the primary somatosensory cortex (S1) and insular,^22,23 and increased activations in PFC and periaqueductal gray matter.⁹ However, no studies have yet investigated functional networks of pain-related brain regions in distraction analgesia. Functional near-infrared spectroscopy (fNIRS) is a non-invasive brain imaging technique primarily used to monitor real-time changes in cortical blood oxygen metabolism to reflect neural activity. Compared to fMRI, fNIRS better accommodates comfortable posture and resists motion artifacts, enabling real-time monitoring of pain responses in this study.²⁴ Time-series analysis of blood-oxygen-dependent signals can capture neural activity of the brain during the process of distraction analgesia modulation. Additionally, by combining neural activity and functional connectivity between brain regions, the cortical regulatory mechanism of cognitive load dependence in distraction analgesia can be deeply understood.²⁵

This study aims to verify the effectiveness of distraction analgesia at both the neural and behavioral levels, investigate the effect of n-back tasks during different WM load on pain perception and to reveal the pattern and cortical mechanism, potentially providing a neuroscientific basis for clinical cognitive-based analgesia interventions.

Material and Methods

Participants

Forty healthy participants (23 females and 17 males; 21–26 years old) were enrolled by social media platforms. Participants had a normal corrected or unaided vision and were right-handed. The exclusion criteria included (1) cardiovascular/respiratory disorders, chronic/acute pain conditions, auditory impairments, or neuropsychiatric diagnoses; (2) pregnancy status, regular drugs use, or chronic medication regimens (excluding contraceptive pills); (3) acute sleep deprivation (<6 hours before the experiment) or recent analgesic/anti-inflammatory drugs administration (<12 hours before the experiment); (4) cognitive and sensory disorders; (5) caffeine intake (<2 hours before the experiment) or intense physical activity on the day of the experiment.

Experimental Design

This experiment employed a mixed design. In the first part of the session, all participants completed a pain calibration followed by a pain-rating task (pain task). In the second part of the session, a 2 × 2 within-subject design was used to assess the distraction effect on pain perception. Participants performed an n-back WM task during two cognitive load conditions: high load (2-back) and low load (0-back), while receiving pain stimuli (with or without laser stimuli) to their right hand. All participants completed tasks in five experimental conditions: pain (laser stimuli without n-back), 0-back (without laser stimuli), 2-back (without laser stimuli), 0-back with pain, and 2-back with pain. Pain intensity ratings and cognitive performance were recorded throughout the experiment.

Experimental Procedures

At the beginning of the study, all participants completed a brief demographic questionnaire. Then, participants underwent a pain calibration procedure. Laser stimuli were delivered to the dorsum of the right hand. After each laser stimulus, participants were instructed to verbally rate the perceived pain intensity using a Numerical Ratings Scale (NRS) ranging from 0 to 10, where 0 represented no pain and 10 represented the most unbearable pain.^26–28 Two different levels of stimuli intensity were determined for each participant, eliciting low (NRS = 4) and high (NRS = 6) pain. After pain calibration, participants were instructed to complete the pain ratings task with fNIRS recording. In the second part of the experiment, participants were required to complete both the n-back task and the n-back with pain task. They were instructed to perform the WM paradigms with fNIRS recording. The n-back task, with different WM load levels (0-back and 2-back), was used to engage WM. Before the task, participants were informed to practice the task and receive real-time feedback of their accuracy to ensure full understanding of the task. During the n-back task with pain (0-back and 2-back tasks with laser stimuli), they completed the n-back task, while pain stimuli were delivered to their dorsum of their right hand. Participants performed the n-back task and the n-back with pain task in a random order. The total duration of the experiment was approximately 50 minutes. The experimental design and procedure are shown in Figure 1.

Figure 1 Experimental design and procedure.

Notes: The numbers inside the squares represent the presentation of random number stimuli. The numbers above the squares indicate the order in which the number stimuli appear (eg 1,3,4). The laser is presented once for every three number stimuli shown.

Pain Task

Noxious laser stimuli (radiant heat) were generated using an infrared Nd:YAP laser (Electronical Engineering, Italy) with a wavelength of 1.34 μm and a pulse duration of 4 milliseconds. The He-Ne laser pulse was transmitted through optical fibers and focused by a lens to a spot approximately 7 mm in diameter,²⁹ synchronously activating nociceptive nerve endings in the superficial skin layers. The noxious stimuli were delivered at 2 individualized energy levels (NRS = 4 and 6) to a circular region (3 cm in diameter) on the dorsum of the right hand. A total of 30 laser stimuli (15 for each intensity) were delivered in a pseudo-randomized sequence, with an interval ranging from 20 to 21 seconds. Participants were uncertain of the exact number of pain stimuli they would experience, and the pain perception induced by laser stimuli was variable. The experiment procedure is shown in Figure 1. Each trial started with a 7–8 second fixation, followed by a short noxious stimulus delivered to the dorsum of the hand. After a 10-second interval, participants were asked to rate the perceived pain intensity using a standardized 0–10 NRS scale. To minimize the risk of nociceptor fatigue or sensitization, the laser target site was manually shifted at least 1 cm in a random direction after each stimulus.

N-Back Task

The WM task used a modified n-back paradigm, with two levels of WM load manipulation (0-back and 2-back). As a validated measure of the central executive system of WM, n-back paradigm reflects core WM functions, including attention control and updating.^30,31 It is widely utilized across psychiatric, neurological, and cognitive research domains.^32,33 The n-back paradigm was designed using E-Prime 3.0 (Psychology Software Tools, Pittsburgh, USA). Random numbers ranging from 0 to 9 were presented on the screen. In the 0-back task, participants were instructed to identify whether the current number was “0” by pressing “Q” for “yes” and “W” for “no”. In the 2-back task, they were required to determine whether the current number matched the one presented before two trials, again pressing “Q” for “yes” and “W” for “no”. The 0-back and 2-back task were repeated in a 0-2-0-2-0-2 sequence in three blocks, with each block containing 30 number stimuli. Thirty percent of the stimuli were target stimuli. The stimuli were presented for 1 second, and participants were given 2 seconds to respond. A 15-second rest interval was implemented between task blocks.

N-Back with Pain Task

The distraction paradigm comprised four conditions in total, each presented in two blocks: 0-back with low laser stimuli, 0-back with high laser stimuli, 2-back with low laser stimuli, and 2-back with high laser stimuli. The order was randomized. The duration of each block was fixed at 60 seconds. Before each block, a cue indicating the upcoming task (0-back or 2-back) was presented for 2 seconds. Each number in the task sequence appeared on the screen for 1000 milliseconds (ms), followed by a 1000 ms blank interval. A total of 10 pain stimuli were delivered during each block. At the end of each block, a 3-second instruction was shown, instructing participants to provide a verbal rating of the perceived pain intensity using a 0–10 NRS.

fNIRS Data Acquisition

In this experiment, a fNIRS system (NirScan, Danyang Huichuang Medical Equipment Co., Ltd., Jiangsu, China) was utilized to assess cortical response from 53 channels. Three different wavelengths (730 nm, 808 nm, and 850 nm) were employed, with a sampling rate of 11 Hz to capture the near-infrared spectroscopy signals. Concentration changes in oxygenated hemoglobin (HbO) and deoxygenated hemoglobin (HbR) were obtained based on the modified Beer-Lambert law. The optical system consisted of 18 sources and 18 detectors, with adjacent sources and detectors spaced 3 cm apart. The connection between sources and detectors was defined as a channel. The coordinates of each probe were determined based on the International 10–20 system and the coordinate localization feature of SPM software, referencing both the coordinates and the optical electrode positions on the Montreal Neurological Institute (MNI) brain template. Figure 2A shows the complete arrangement of the fNIRS probes and channels, while Figure 2B shows the location of the 41 selected channels out of 53, covering 9 brain regions of interest: right and left secondary somatosensory cortex (RS2; LS2), right and left premotor cortex (RPMC; LPMC), right and left primary somatosensory cortex (RSM1; LSM1), right and left dorsolateral prefrontal cortex (RDLPFC; LDLPFC), and anterior prefrontal cortex (aPFC).

Figure 2 fNIRS channel settings.

Abbreviations: aPFC, anterior prefrontal cortex; DLPFC, dorsolateral prefrontal cortex; PMC, premotor cortex; SM1, primary somatosensory cortex; S2, secondary somatosensory cortex; L, left; R, right; S, source; D, detector.

Notes: fNIRS 36-probe and 53-channel montage placement (A) and the distribution of fNIRS 53-channel in S2, PMC, SM1, DLPFC and aPFC (B). The numbers in (A) represent 18 light sources and detectors. The numbers in (B) represent 53 channels.

fNIRS Preprocessing and Analysis

In this study, fNIRS signals were preprocessed and analyzed using NirSpark software (HuiChuang, China) and matlab (Mathworks, MA, USA). First, we conducted a preliminary inspection of the raw data, identifying and removing channels with poor signal quality. Signal quality was assessed based on the coefficient of variation (cv), where values ≤5 were considered good, 5–20 were considered acceptable, and >20 were considered poor. Next, the raw light intensity data series were converted into optical density (OD) changes. We applied spline interpolation to correct motion artifacts in the channels. Motion artifacts typically manifest as pulsatile or abrupt jumps caused by relative movement between the scalp and the probe.³⁴ We applied a bandpass filter (0.01–0.2 Hz) to remove physiological noise, such as respiration, cardiac activity, and low-frequency signal drift. Subsequently, based on the modified Beer-Lambert law, we calculated changes in oxygenated hemoglobin (HbO), deoxygenated hemoglobin (HbR), and total hemoglobin (HbT) concentrations.

During the analysis phase, we focused on data from 41 channels that covered nine ROIs, which are significantly associated with pain based on previous studies. Specifically, multiple studies have demonstrated that brain regions such as S1, S2, M1, aPFC, DLPFC and PMC are critical for pain processing.³⁵ Previous studies on distraction analgesia have confirmed that during the process of distraction, activation in brain regions such as the insula and S1 decreases, while neural responses in regions including the DLPFC and parietal lobe show increased activation and reduced functional connectivity.³⁶ Meanwhile, brain function studies related to analgesia have also reported that when pain perception decreases, the activation level of SM1 and its functional connectivity with the DLPFC/S1 are significantly reduced.³⁷ Different time windows were set to extract stable hemoglobin time series: for the pain task, the time window was 13 seconds after each stimulus; for the n-back task, the time window was 30 seconds; and for the n-back with pain task, the time window was 60 seconds. The 2-second pre-task period was used as a baseline for correction. By averaging across all blocks for each task, we generated the mean hemodynamic reaction time series curves related to the task. Previous studies have shown that HbO is more sensitive to changes in brain region blood flow signals than HbR,^38,39 therefore, we primarily focused on the changes in the mean HbO values under different task conditions as indicators of brain activation.

To further explore the load-dependent neural mechanism of distraction analgesia, we conducted functional connectivity analysis to observe the inter-regional connectivity of the brain during different tasks. The relative changes in HbO concentration within each block were extracted for functional connectivity analysis using the brain network module in the NirSpark software package. Pearson correlation coefficients between the HbO concentrations of different brain regions in the time series were calculated, followed by Fisher Z transformation. The transformed values were then defined as the functional connectivity strength.

Statistical Analysis

Data was analyzed using IBM SPSS Statistics for Windows, Version 25.0 (Armonk, NY: IBM Corp). The measured data were expressed as means and standard deviations. Normality of the data was assessed using the Shapiro–Wilk test. Independent samples t-tests and paired t-tests were applied for normally distributed continuous variables, while non-parametric tests were used for variables that were not normally distributed. To compare the effect of different WM loads on pain ratings, a one-way repeated measures analysis of variance (ANOVA) was conducted, with false discovery rate (FDR) correction applied. At the neural level, we first verified whether pain elicited activation in the corresponding brain regions by conducting paired t-tests on the HbO mean values for the n-back task and n-back with pain task, with FDR correction applied. Secondly, we analyzed the impact of WM load on pain-related neural activity by conducting paired t-tests on the HbO mean values for 0-back with pain task and 2-back with pain task, with FDR correction applied. To examine whether WM load influenced pain-related behavior through alterations in pain-related neural activity, a mediation analysis was conducted by using the SPSS 25 (test of joint significance approach) and Mplus8.11 (path-analytic method).⁴⁰ In this model, the independent variable (X) represented the WM load (0-back = 1; 2-back = −1), the dependent variable (Y) was the pain-related behavioral measures (defined as the change in pain ratings on the NRS), and the mediator (M) was the pain-related neural activity. Pain-related neural activity and behavioral measures were measured twice in the same subject at both WM loads. The pain-related behavioral or neural measures were quantified as the contrast between the n-back with pain task and n-back task. They were measured twice in the same subject at both WM loads. The indirect effect was considered statistically significant when the 95% confidence interval (CI) did not include zero, with a significance threshold set at p < 0.05. Additionally, a two-way repeated measures ANOVA was used to compare the differences in functional connectivity between brain regions across the n-back and n-back with pain tasks, and simple effects analysis was conducted when the interaction effect was significant. To assess the interference of pain on cognitive task performance, a two-way repeated measures ANOVA was conducted to compare WM reaction time (RT) and accuracy (ACC) for each task (0-back and 2-back) when performed with or without pain. Simple effects analysis was conducted when the interaction effect was significant, with FDR applied. The partial eta squared (η²p) effect sizes of significant effects in the ANOVA were reported, where 0.01 represented a small effect, 0.06 represented a medium effect, and 0.14 represented a large effect.⁴¹ Statistical significance was set at p < 0.05 for all tests, and all p – values were two-tailed.

Power Analysis

The primary outcome of this study was the effect of WM load on pain ratings, which was assessed using a one-way repeated measures ANOVA. An a priori power analysis conducted with G*Power 3.1, assuming a medium effect size (f = 0.25), a significance level of α = 0.05, and a statistical power of (1 – β) = 0.80, indicated a required sample size (n = 28), ensuring adequate power at the behavioral level. To ensure sufficient power for neural and connectivity analyses, additional power calculations were performed for paired-sample t-tests and two-way repeated measures ANOVA. The sample size are n = 27 and n = 24. The largest sample size among these analyses was adopted as the reference (n = 28). Considering a 10–15% attrition rate, a minimum of 32 participants should be enrolled in the study. Thus, the study was sufficiently powered across behavioral, neural, and connectivity measures.

Results

Descriptive Statistics

Five participants were excluded from the initial sample of N = 40. Data from 2 participants were incomplete due to missing marking records caused by equipment failure. In addition, data from 3 participants were excluded due to excessive motion during the task phase of fNIRS measurements (n = 2) and poor signal quality in some channels caused by inadequate fitting of the measurement cap to the head shape (n = 1). Thus, we analyzed data from 35 participants. Table 1 presents the baseline demographic and pain-related clinical characteristics of the participants. The gender ratio of participants included in this study was not balanced. Results from baseline-related questionnaires have excluded the confounding effect of gender, confirming the homogeneity and stability of the study sample.

Table 1 Demographic and Pain-Related Clinical Characteristics of the Participants

Modulation of Pain Ratings by WM Load

As shown in Figure 3, Repeated measures ANOVA was performed on the perceived pain intensity NRS ratings during different conditions (pain, 0-back with pain, and 2-back with pain). The results showed significant differences in pain intensity ratings across conditions (F = 17.666; p < 0.001***; η²p = 0.342). Compared to the pain task, NRS were significantly lower in both 0-back with pain (p = 0.001**) and 2-back with pain (p < 0.001***). Additionally, NRS was significantly lower in the 2-back with pain compared to 0-back with pain (p = 0.002**).

Figure 3 Pain ratings in pain task and n-back with pain task.

Notes: Violin plots show the data distributions, mean (dashed line), and quartiles (black line). Each colored dot represents an individual participant. ***p < 0.001.

Modulation of Pain-Related Brain Activity by WM Load

In the whole-brain analysis, noxious laser stimuli activated a broad range of brain areas associated with pain, including RS2 (t = 2.473, p = 0.034*), RPMC (t = 3.108, p = 0.012*), RSM1 (t = 2.210, p = 0.044*), LSM1 (t = 2.262, p = 0.044*), RDLPFC (t = 3.932, p < 0.001***), LDLPFC (t = 2.716, p = 0.023*), and aPFC (t = 3.655, p = 0.005**), as shown in Table 2 and Figure 4.

Table 2 Effect of N-Back with Pain Task on Brain Responses to Pain Stimuli

Figure 4 Brain activation during the n-back and n-back with pain task.

Abbreviations: aPFC, anterior prefrontal cortex; DLPFC, dorsolateral prefrontal cortex; PMC, premotor cortex; SM1, primary somatosensory cortex; S2, secondary somatosensory cortex; L, left; R, right.

Notes: Brain activation plots for n-back (A), n-back with pain (B) and the difference (C). The redder the color in the brain map, the greater the activation.

Further comparison of brain activation during 0-back with pain and 2-back with pain task revealed significant decreases in activation in the left S2 (t = 2.757, p = 0.041*) and left SM1 (t = 2.834, p = 0.041*) regions during 2-back with pain, as shown in Table 3 and Figure 5. Mediation analysis was used to determine the contribution of neural activity in the effect of cognitive load on pain perception. As shown in Figure 6, WM load indirectly affected pain intensity by modulating brain activity in the LSM1 (a*b = −0.014, SE = 0.039, CI = [−0.102, 0.059]) and in the LS2 (a*b = 0.004, SE = 0.050, CI = [−0.112, 0.096]), but the effects were not significant. The point estimates of the indirect effects were not close to zero. The result is more likely attributed to insufficient statistical power (limited sample size) rather than a genuine absence of the mediating effect. Thus, the results could not confirm the mediating role of LSM1/LS2 neural activity, but they also do not rule out this potential pathway. A two-way repeated measures ANOVA was used to compare the functional connectivity between brain regions during different tasks. Significant interaction effects were found in the functional connectivity between RS2-aPFC (F = 6.475, p = 0.016*, η²p = 0.160), RSM1-RDLPFC (F = 6.225, p = 0.018*, η²p = 0.155), RSM1-aPFC (F = 7.439, p = 0.010**, η²p = 0.180), and LSM1-aPFC (F = 6.523, p = 0.015*, η²p = 0.161), as shown in Table 4. Simple effects analysis was conducted on the significant interaction effects in the functional connectivity between brain regions.

Table 3 Brain Responses to N-Back with Pain Task with Different WM Load

Table 4 Effect of WM Load and Pain Distraction on Brain Functional Connectivity

Figure 5 Brain activation differences during the n-back with pain task during different WM load in LS2 and LSM1.

Abbreviations: LSM1, left primary somatosensory cortex; LS2, left secondary somatosensory cortex.

Notes: Violin plots and brain activation plots for LS2 (A) and the LSM1 (B). Violin plots show the data distributions, mean (dashed line), and quartiles (black line). Each colored dot represents an individual participant. Positive values represent positive activation (increase), negative values represent negative activation (decrease). The redder the color in the brain map, the greater the activation. *p < 0.05.

Figure 6 Mediating role of neural responses on the effect that WM had on pain perception.

Abbreviations: LSM1, left primary somatosensory cortex; LS2, left secondary somatosensory cortex. SE, standard error.

Notes: Mediating role of LSM1 neural responses (A) and LS2 neural responses (B) on the effect that WM had on pain perception. Violin plots show the data distributions, mean (dashed line), and quartiles (black line). Each colored dot represents an individual participant. The redder the color in the brain map, the greater the activation. Dotted paths indicate significance, while solid paths indicate non-significance. *p < 0.05; **p < 0.01; ***p < 0.001.

The results revealed that during high load task, the additional pain stimuli in the n-back with pain task reduced the functional connectivity between brain regions compared to the n-back task in RS2-aPFC (p = 0.003**), RSM1-RDLPFC (p < 0.001***), RSM1-aPFC (p = 0.004**), and LSM1-aPFC (p = 0.034*), as shown in Figure 7A. Functional connectivity in the n-back task increased in RS2-aPFC (p = 0.002**), RSM1-RDLPFC (p = 0.002**), and RSM1-aPFC (p = 0.003**) with increasing load, as shown in Figure 7B.

Figure 7 Results of functional connectivity differences within the high load WM and the n-back.

Abbreviations: aPFC, anterior prefrontal cortex; DLPFC, dorsolateral prefrontal cortex; SM1, primary somatosensory cortex; S2, secondary somatosensory cortex; L, left; R, right.

Notes: Violin plots and FC plots for the high load WM (A) and the n-back (B). Violin plots show the data distributions, mean (dashed line), and quartiles (black line). Each colored dot represents an individual participant. The redder the color in the line of the brain map, the greater the differences. *p < 0.05; **p < 0.01; ***p < 0.001.

Interference of Pain with Cognitive Performance

A two-way repeated measures ANOVA was conducted to compare WM RT and ACC for each task (0-back and 2-back) when performed with or without pain. No significant interaction effects were found between WM load and pain intensity on ACC (F = 0.628, p = 0.434) and RT (F = 0.0005, p = 0.983). Focusing on the main effects, significant main effects of WM load were found on both ACC (F = 28.799, p < 0.001***, η²p = 0.459) and RT (F = 61.253, p < 0.001***, η²p = 0.643). Additionally, a significant main effect of pain was found in ACC (F = 5.346, p = 0.027*, η²p = 0.136), as shown in Figure 8.

Figure 8 Reaction time and accuracy during different WM load.

Abbreviations: ACC, accuracy; RT, reaction time.

Notes: Reaction time during different WM load (A) and accuracy during different WM load (B). Violin plots show the data distributions, mean (dashed line), and quartiles (black line). Each colored dot represents an individual participant. *p < 0.05; ***p < 0.001.

Discussion

This study used fNIRS to explore cognitive load-dependent perception modulation and cortical mechanism in distraction analgesia in healthy individuals. The results showed that high-load WM significantly reduced the perceived intensity⁴² and pain-related neural activation in the S2 and SM1. Under high load, the functional connectivity between brain regions (RS2-aPFC, RSM1-RDLPFC, RSM1-aPFC, and LSM1-aPFC) was significantly lower during the n-back with pain task compared to the n-back task. It indicated that as WM load increased, the coupling between the network involved in pain processing was significantly attenuated.

Based on the neurocognitive model of pain, WM regulates the allocation of attention resources through central executive system, and its load levels directly affect the effectiveness of pain perception suppression.⁴³ This study systematically revealed the gradient effect of WM load on pain perception inhibition by manipulating load levels of WM. We found that WM significantly reduced pain ratings, with the analgesic effect of the 2-back task being greater than that of the 0-back task. Our findings were consistent with prior research,²⁰ which involves shifting cognitive resources away from pain and prioritizing task-related stimuli through “top-down” control. This reduces the occupation of limited cognitive resources by pain, thereby decreasing pain perception.⁷ Furthermore, considering that pain expectation and continuous stimulation may lead to a decline in perceptual levels,⁴⁴ we set two different pain levels and randomized both the spatiotemporal delivery (arrival and location) and intensity levels of noxious stimuli. This setting effectively avoided interference from expectation effects and sensory habituation during pain assessment,⁴⁵ reinforcing the central role of the resource competition theory. Although this study found a positive correlation between WM load and pain suppression, previous research suggests that this modulation pattern may be influenced by factors such as cognitive fatigue and ceiling effects.²⁰ A pain study by Zoha Deldar et al²⁰ found that both 0-back and n-back tasks reduced pain, with n-back being more effective than 0-back, but no significant differences in pain suppression were observed between 2-back and 3-back tasks. Based on the research by Vogel et al²¹, we proposed potential explanations for the absent analgesic effect in n-back tasks. WM affects the allocation of attention resources in competition, when load exceeds individuals’ execution capacity threshold, indirectly reversing the analgesic advantage and reducing the contribution of WM to distraction analgesia. Therefore, the ceiling effect caused by limited WM capacity is one of the important factors affecting distraction analgesia. Future research can establish a multi-gradient model of WM capacity and assess individual difference in execution function to explore the optimal load range for maximizing analgesic effects.

At the neural level, analgesia effects induced by WM distraction might involve a hierarchical gate control mechanism driven by PFC. We compared brain activation between the n-back task and the n-back with pain task, and found that SM1, S2, PMC, aPFC, and DLPFC were activated during pain stimuli in the n-back with pain task, forming a dynamic regulatory network for pain processing.³⁵ SM1 and S2 are involved in nociceptive pain and sensory processing,^46,47 and more significant brain activation during 0-back suggests less inhibitory effect of low WM load on pain processing. As WM load increased, activation in the S2 and SM1 decreased during the n-back with pain task. We supposed that high WM load might enhance thalamocortical inhibitory gating, resulting in suppression of nociceptive signal transmission from S2 and SM1. An fMRI study by Valet et al⁴⁸ showed that attention diversion induced by single cognitive load Stroop tasks reduced brain activities encoding pain, such as the thalamus and somatosensory cortex, supporting the effect of distraction analgesia at a single-load level. Moreover, an fMRI study by Legrain V et al^49,50 found that when focusing on a primary visual task, the responses in S1/M1 and the insula to noxious stimuli are reduced. Neural responses in S1/S2 are affected by attention loads, with the spatial pattern of distraction analgesia being highly consistent with the findings of this study. Based on fNIRS data, this study further suggests that WM distraction modulates pain processing through “top-down” attention regulation, reducing somatosensory cortical neural activity and thus affecting the processing of pain stimuli.

We conducted mediation analysis on the neural responses of the SM1 and S2 regions to test whether WM loads indirectly affected pain ratings through brain activation. The results showed no significant mediation effect, indicating that the effect of loads on pain ratings was not completely dependent on the neural response changes in individual brain regions. The effect of distraction analgesia may involve a complex functional network. Therefore, to observe the changes in brain functional networks induced by distraction under different loads, we calculated the functional connectivity between target brain regions.

The results of the functional connectivity between target brain regions indicated that with increasing load, functional connectivity in the WM task increased between RS2-aPFC, RSM1-aPFC, and RDLPFC-RSM1, reflecting the recruitment of “top-down” modulatory resources at the cortical level due to WM load. However, as the load increased, the additional pain stimuli in the n-back with pain task reduced the functional connectivity between the RS2-aPFC, bilateral SM1-aPFC, and RDLPFC-RSM1 brain regions compared to the n-back task. The results indicated that high load distraction inhibited the control of the sensory cortex by the PFC and reduced the coupling between pain-related brain networks, which might be related to the saturation effect in prefrontal control pathways. The DLPFC supports executive control over attention, WM, and pain inhibition via top-down modulation.^51,52 aPFC, an essential component of the PFC, contributes to self-referential processing, attention regulation, WM, decision-making, and salience detection. These findings indicated that aPFC was important in appropriate attention shift and the reallocation of pain awareness and response.³⁵ The functional connectivity between the PFC and SM1 might indicate the contribution of prefrontal cognitive processing to pain processing. The PFC might be a key node in the network related to nociceptive processing and pain regulation, transmitting core pain processing through its connection with SM1.⁵² A study by Peng Weiwei et al³⁷ found that α-tACS on SM1 may inhibit pain perception and neural responses by decoupling SM1 from key sensory-motor, emotional, and cognitive processing networks involved in pain. Similar findings were also reported by Wagner et al in their study on placebo analgesia.⁵³ They found that placebo manipulation may exert analgesic effects by decoupling the somatosensory network responsible for pain processing from the descending modulatory network. Furthermore, Deng xue et al^54,55 reported that VR-induced analgesia, as shown in fNIRS studies, is characterized by reduced S1 connectivity and diminished pain-related processing. The weakened coupling between cortical pain-related regions and other brain areas may serve as a critical mechanism disrupting normal pain signaling, consistent with our findings. Therefore, we speculated that WM distraction analgesia may similarly reduce pain through decoupling the core networks involved in pain perception and cognitive processing.

We also examined the potential impact of pain on WM performance. A significant main effect of pain intensity on ACC was observed, indicating that the presence of pain significantly impaired WM performance. This damage might arise through two potential pathways. Pain signals are transmitted through the spinal-thalamic-cortical pathway to cortical regions such as S1 and S2.⁵⁶ Through inter-regional neural connectivity or reorganization, the PFC evaluates pain and allocates attention, thereby competing for cognitive resources and reducing the neural encoding precision of WM representations.⁵⁷ Additionally, the insula and DLPFC, as key regions for cognitive control, may be involved in analgesia processing. This mechanism might relate to diminished “top-down” cognitive control from the DLPFC to the insula, ACC, and thalamus.^58–61 Pain-induced negative emotions may decrease the coupling of the PFC and ACC during cognitive reappraisal.²² Moreover, there was no significant interaction between WM load and pain in terms of accuracy, suggesting that the impairing effect of pain might be similar under both 0-back and 2-back conditions. In high-load condition, the increase in reaction time alongside a decrease in accuracy supports Baddeley’s limited capacity theory of the central executive system.⁶² When cognitive demand exceeds the individual’s resource threshold, the cost of conflict monitoring in the ventral attention network (VAN) increases, and more time is needed for information matching and conflict resolution.⁶³ Notably, the observed “speed-accuracy decoupling” under pain conditions suggests the adoption of a behavioral compensation strategy, whereby participants maintain response speed at the expense of accuracy.⁶⁴ Our findings support the hierarchical hypothesis of the attention competition model, demonstrating that pain disrupts cognition through two mechanisms: (1) direct competition for prefrontal resources, and (2) emotion-mediated impairment of cognitive control networks.²² This provides a new interpretation for cognitive deficits observed in chronic pain. Prolonged pain may reconstruct neural networks, reduce the availability of cognitive resources and reinforce negative affective processing in the limbic system,^65,66 leading to a vicious “pain-cognition” cycle and driving the brain’s functional reorganization toward a “pain salience-prioritized” mode. Future research may focus on interventions such as neurofeedback training, transcranial electrical stimulation, or virtual reality-based distraction tasks to reconstruct neural connectivity, and improve cognitive impairments in chronic pain patients.

The results of the current study should be considered within the study design and its potential limitations. First, as the participants were all young and healthy university students, the findings may not be generalizable to broader or more diverse populations. Second, the load of the WM paradigm was simple. Future research could use more refined gradients and incorporate broader attention tests. Third, the indirect effects of WM load on pain ratings via LSM1/LS2 did not reach significance, which may be attributed to insufficient statistical power. Future research could expand the sample size and improve measurement precision.

Conclusion

Our study demonstrates that WM distraction reduces both pain perception and neural responses to experimental laser pain stimuli in healthy individuals and that a significant reduction in functional coupling between regional networks involved in pain processing was observed. Therefore, our findings suggest that WM load may reduce pain perception by decreasing neural responses in pain-related regions and promoting the decoupling of related brain networks, providing neuroscientific evidence for cognitive strategy-based analgesia interventions.

Data Sharing Statement

For additional details, please reach out to the corresponding author. The datasets analyzed in this study can be obtained from the corresponding author upon reasonable request.

Ethics Approval and Informed Consent

All participants gave written informed consent. The study procedures were approved by the Ethics Committee of Zhujiang Hospital, Southern Medical University (Ethical approval number: 2024-KY-427-02). The clinical registration number is ChiCTR2500100508. This study complies with the Declaration of Helsinki.

Author Contributions

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Funding

This study was supported by National Natural Science Foundation of China (NNSFC), China; Contract grant number: 82172526, 82372553; Guangdong Basic and Applied Basic Research Foundation, China; Contract grant number: 2023A1515010200.

Disclosure

The author(s) report no conflicts of interest in this work.

References

1. Broadbent P, Liossi C, Schoth DE. Attentional bias to somatosensory stimuli in chronic pain patients: a systematic review and meta-analysis. Pain. 2021;162(2):332–352. doi:10.1097/j.pain.0000000000002040

2. Abudoush AN, Noureen A, Panagioti M, et al. What can we learn about selective attention processes in individuals with chronic pain using reaction time tasks? A systematic review and meta-analysis. Pain. 2023;164(8):1677–1692. doi:10.1097/j.pain.0000000000002885

3. Legrain V, Mancini F, Sambo CF, Torta DM, Ronga I, Valentini E. Cognitive aspects of nociception and pain: bridging neurophysiology with cognitive psychology. Clinical Neurophysiology. 2012;42(5):325–336. doi:10.1016/j.neucli.2012.06.003

4. Ryckeghem D, Crombez G. Pain and Attention: Toward a Motivational Account. Motivational Perspectives on Chronic Pain. Oxford University Press; 2018.

5. Van Ryckeghem DM, Crombez G, Eccleston C, Legrain V, Van Damme S. Keeping pain out of your mind: the role of attentional set in pain. European Journal of Pain. 2013;17(3):402–411. doi:10.1002/j.1532-2149.2012.00195.x

6. Fisher JT, Keene JR, Huskey R, Weber R. The Limited Capacity Model of Motivated Mediated Message Processing: taking Stock of the Past. Annals of the International Communication Association. 2018;42(4):270–290. doi:10.1080/23808985.2018.1534552

7. Phelps CE, Navratilova E, Porreca F. Cognition in the Chronic Pain Experience: preclinical Insights. Trends in Cognitive Sciences. 2021;25(5):365–376. doi:10.1016/j.tics.2021.01.001

8. Torta DM, Legrain V, Mouraux A, Valentini E. Attention to pain! A neurocognitive perspective on attentional modulation of pain in neuroimaging studies. Cortex. 2017;89:120–134. doi:10.1016/j.cortex.2017.01.010

9. Birnie KA, Chambers CT, Spellman CM. Mechanisms of distraction in acute pain perception and modulation. Pain. 2017;158(6):1012–1013. doi:10.1097/j.pain.0000000000000913

10. Torta DM, De Laurentis M, Eichin KN, von Leupoldt A, van den Broeke EN, Vlaeyen JWS. A highly cognitive demanding working memory task may prevent the development of nociceptive hypersensitivity. Pain. 2020;161(7):1459–1469. doi:10.1097/j.pain.0000000000001841

11. Legrain V, Crombez G, Plaghki L, Mouraux A. Shielding cognition from nociception with working memory. Cortex. 2013;49(7):1922–1934. doi:10.1016/j.cortex.2012.08.014

12. Awh E, Vogel EK, Oh SH. Interactions between attention and working memory. Neuroscience. 2006;139(1):201–208. doi:10.1016/j.neuroscience.2005.08.023

13. D’Esposito M, Postle BR. The cognitive neuroscience of working memory. Annual Review of Psychology. 2015;66(1):115–142. doi:10.1146/annurev-psych-010814-015031

14. Gazzaley A, Nobre AC. Top-down modulation: bridging selective attention and working memory. Trends in Cognitive Sciences. 2012;16(2):129–135. doi:10.1016/j.tics.2011.11.014

15. Postle BR, Awh E, Jonides J, Smith EE, D’Esposito M. The where and how of attention-based rehearsal in spatial working memory. Brain Research Cognitive Brain Research. 2004;20(2):194–205. doi:10.1016/j.cogbrainres.2004.02.008

16. SanMiguel I, Corral MJ, Escera C. When loading working memory reduces distraction: behavioral and electrophysiological evidence from an auditory-visual distraction paradigm. Journal of Cognitive Neuroscience. 2008;20(7):1131–1145. doi:10.1162/jocn.2008.20078

17. Morogiello J, Murray NG, Hunt TN, Harris BS, Szekely BJ, Shaver GW. The effect of acute pain on executive function. Journal of Clinical and Translational Research. 2019;4(2):113–121.

18. Buhle J, Wager TD. Performance-dependent inhibition of pain by an executive working memory task. Pain. 2010;149(1):19–26. doi:10.1016/j.pain.2009.10.027

19. Legrain V, Bruyer R, Guérit JM, Plaghki L. Involuntary orientation of attention to unattended deviant nociceptive stimuli is modulated by concomitant visual task difficulty. Evidence from laser evoked potentials. Clinical Neurophysiology: Official Journal of the International Federation of Clinical Neurophysiology. 2005;116(9):2165–2174. doi:10.1016/j.clinph.2005.05.019

20. Deldar Z, Blanchette I, Piché M. Reduction of Pain and Spinal Nociceptive Transmission by Working Memory is Load Dependant. The Journal of Pain. 2021;22(7):797–805. doi:10.1016/j.jpain.2021.02.001

21. Vogel TA, Desjardins S, Chabot B, Otto AR, Falk CF, Roy M. Cognitive fatigue weakens the effects of distraction on pain. Pain. 2023;164(12):2845–2851. doi:10.1097/j.pain.0000000000002985

22. Bushnell MC, Ceko M, Low LA. Cognitive and emotional control of pain and its disruption in chronic pain. Nature Reviews Neuroscience. 2013;14(7):502–511. doi:10.1038/nrn3516

23. Wiech K, Ploner M, Tracey I. Neurocognitive aspects of pain perception. Trends in Cognitive Sciences. 2008;12(8):306–313. doi:10.1016/j.tics.2008.05.005

24. Cheng X, Guo B, Hu Y. Distinct neural couplings to shared goal and action coordination in joint action: evidence based on fNIRS hyperscanning. Social Cognitive and Affective Neuroscience. 2022;17(10):956–964. doi:10.1093/scan/nsac022

25. Hu XS, Nascimento TD, DaSilva AF. Shedding light on pain for the clinic: a comprehensive review of using functional near-infrared spectroscopy to monitor its process in the brain. Pain. 2021;162(12):2805–2820. doi:10.1097/j.pain.0000000000002293

26. Ferreira-Valente MA, Pais-Ribeiro JL, Jensen MP. Validity of four pain intensity rating scales. Pain. 2011;152(10):2399–2404. doi:10.1016/j.pain.2011.07.005

27. Hawker GA, Mian S, Kendzerska T, French M. Measures of adult pain: visual Analog Scale for Pain (VAS Pain), Numeric Rating Scale for Pain (NRS Pain), McGill Pain Questionnaire (MPQ), Short-Form McGill Pain Questionnaire (SF-MPQ), Chronic Pain Grade Scale (CPGS), Short Form-36 Bodily Pain Scale (SF-36 BPS), and Measure of Intermittent and Constant Osteoarthritis Pain (ICOAP). Arthritis Care & Research. 2011;63(11):S240–52.

28. Travers M, Wand BM, Hince D, et al. The Effect of Information on the Time Course of Pain During an Episode of Acute Experimentally Induced Low Back Pain-A Randomised Experiment. European Journal of Pain. 2025;29(5):e70011. doi:10.1002/ejp.70011

29. Hu L, Cai MM, Xiao P, Luo F, Iannetti GD. Human brain responses to concomitant stimulation of Aδ and C nociceptors. The Journal of Neuroscience: the Official Journal of the Society for Neuroscience. 2014;34(34):11439–11451. doi:10.1523/JNEUROSCI.1355-14.2014

30. Pahor A, Seitz AR, Jaeggi SM. Near transfer to an unrelated N-back task mediates the effect of N-back working memory training on matrix reasoning. Nature Human Behaviour. 2022;6(9):1243–1256. doi:10.1038/s41562-022-01384-w

31. Zhou H, Wu J, Li J, et al. Event cache: an independent component in working memory. Science Advances. 2025;11(21):eadt3063. doi:10.1126/sciadv.adt3063

32. Yaple Z, Arsalidou M. N-back Working Memory Task: meta-analysis of Normative fMRI Studies With Children. Child Development. 2018;89(6):2010–2022. doi:10.1111/cdev.13080

33. Nikolin S, Tan YY, Schwaab A, Moffa A, Loo CK, Martin D. An investigation of working memory deficits in depression using the n-back task: a systematic review and meta-analysis. Journal of Affective Disorders. Apr. 2021;284:1–8. doi:10.1016/j.jad.2021.01.084

34. Sutoko S, Monden Y, Tokuda T, et al. Atypical Dynamic-Connectivity Recruitment in Attention-Deficit/Hyperactivity Disorder Children: an Insight Into Task-Based Dynamic Connectivity Through an fNIRS Study. Frontiers in Human Neuroscience. 2020;14:3. doi:10.3389/fnhum.2020.00003

35. Karunakaran KD, Peng K, Berry D, et al. NIRS measures in pain and analgesia: fundamentals, features, and function. Neuroscience and Biobehavioral Reviews. 2021;120:335–353. doi:10.1016/j.neubiorev.2020.10.023

36. Silvestrini N, Corradi-Dell’Acqua C. Distraction and cognitive control independently impact parietal and prefrontal response to pain. Social Cognitive and Affective Neuroscience. 2023;18(1). doi:10.1093/scan/nsad018

37. Peng W, Zhan Y, Jin R, Lou W, Li X. Aftereffects of alpha transcranial alternating current stimulation over the primary sensorimotor cortex on cortical processing of pain. Pain. 2023;164(6):1280–1290. doi:10.1097/j.pain.0000000000002814

38. Hoshi Y, Kobayashi N, Tamura M. Interpretation of near-infrared spectroscopy signals: a study with a newly developed perfused rat brain model. Journal of applied Physiology. 2001;90(5):1657–1662. doi:10.1152/jappl.2001.90.5.1657

39. Strangman G, Culver JP, Thompson JH, Boas DA. A quantitative comparison of simultaneous BOLD fMRI and NIRS recordings during functional brain activation. Neuroimage. 2002;17(2):719–731. doi:10.1006/nimg.2002.1227

40. Wang Y, Wen Z. The Analyses of Mediation Effects Based on Two-Condition Within-Participant Design. Journal of Psychological Science. 2018;41(5):1233–1239.

41. Richardson JTE. Eta squared and partial eta squared as measures of effect size in educational research. Educational Research Review. 2011;6(2):135–147. doi:10.1016/j.edurev.2010.12.001

42. Rischer KM, González-Roldán AM, Montoya P, Gigl S, Anton F, van der Meulen M. Distraction from pain: the role of selective attention and pain catastrophizing. European Journal of Pain. 2020;24(10):1880–1891. doi:10.1002/ejp.1634

43. Legrain V, Damme SV, Eccleston C, Davis KD, Seminowicz DA, Crombez G. A neurocognitive model of attention to pain: behavioral and neuroimaging evidence. Pain. 2009;144(3):230–232. doi:10.1016/j.pain.2009.03.020

44. Seidel EM, Pfabigan DM, Hahn A, et al. Uncertainty during pain anticipation: the adaptive value of preparatory processes. Human Brain Mapping. 2015;36(2):744–755. doi:10.1002/hbm.22661

45. Habermann M, Strube A, Büchel C. How control modulates pain. Trends in Cognitive Sciences. 2025;29(1):60–72. doi:10.1016/j.tics.2024.09.014

46. Canavero S, Bonicalzi V. Role of primary somatosensory cortex in the coding of pain. Pain. 2013;154(7):1156–1158. doi:10.1016/j.pain.2013.02.032

47. Muthalib M, Re R, Zucchelli L, et al. Effects of Increasing Neuromuscular Electrical Stimulation Current Intensity on Cortical Sensorimotor Network Activation: a Time Domain fNIRS Study. PLoS One. 2015;10(7):e0131951. doi:10.1371/journal.pone.0131951

48. Valet M, Sprenger T, Boecker H, et al. Distraction modulates connectivity of the cingulo-frontal cortex and the midbrain during pain–an fMRI analysis. Pain. 2004;109(3):399–408. doi:10.1016/j.pain.2004.02.033

49. Seminowicz DA, Davis KD. Interactions of pain intensity and cognitive load: the brain stays on task. Cerebral Cortex. 2007;17(6):1412–1422. doi:10.1093/cercor/bhl052

50. Seminowicz DA, Davis KD. Pain enhances functional connectivity of a brain network evoked by performance of a cognitive task. Journal of Neurophysiology. 2007;97(5):3651–3659. doi:10.1152/jn.01210.2006

51. Apkarian AV, Sosa Y, Sonty S, et al. Chronic back pain is associated with decreased prefrontal and thalamic gray matter density. The Journal of Neuroscience: the Official Journal of the Society for Neuroscience. 2004;24(46):10410–10415. doi:10.1523/JNEUROSCI.2541-04.2004

52. Seminowicz DA, Moayedi M. The Dorsolateral Prefrontal Cortex in Acute and Chronic Pain. The Journal of Pain. 2017;18(9):1027–1035. doi:10.1016/j.jpain.2017.03.008

53. Wagner IC, Rütgen M, Hummer A, Windischberger C, Lamm C. Placebo-induced pain reduction is associated with negative coupling between brain networks at rest. Neuroimage. 2020;219:117024. doi:10.1016/j.neuroimage.2020.117024

54. Deng X, Jian C, Yang Q, Jiang N, Huang Z, Zhao S. The analgesic effect of different interactive modes of virtual reality: a prospective functional near-infrared spectroscopy (fNIRS) study. Frontiers in Neuroscience. 2022;16:1033155. doi:10.3389/fnins.2022.1033155

55. Deng X, Jiang N, Huang Z, Wang Q. Cortical Activation and Functional Connectivity Response to Different Interactive Modes of Virtual Reality (VR)-Induced Analgesia: a Prospective Functional Near-Infrared Spectroscopy (fNIRS) Study. Journal of Pain Research. 2025;18:1095–1108. doi:10.2147/JPR.S492361

56. Coghill RC. The Distributed Nociceptive System: a Framework for Understanding Pain. Trends in Neurosciences. 2020;43(10):780–794. doi:10.1016/j.tins.2020.07.004

57. Liang HB, He WY, Liu YP, Wang HB. Pain Comorbidities with Attention Deficit: a Narrative Review of Clinical and Preclinical Research. Journal of Pain Research. 2024;17:1055–1065. doi:10.2147/JPR.S443915

58. Uddin LQ. Salience processing and insular cortical function and dysfunction. Nature Reviews Neuroscience. 2015;16(1):55–61. doi:10.1038/nrn3857

59. Cai W, Ryali S, Chen T, Li CS, Menon V. Dissociable roles of right inferior frontal cortex and anterior insula in inhibitory control: evidence from intrinsic and task-related functional parcellation, connectivity, and response profile analyses across multiple datasets. The Journal of Neuroscience: the Official Journal of the Society for Neuroscience. 2014;34(44):14652–14667. doi:10.1523/JNEUROSCI.3048-14.2014

60. Corbetta M, Shulman GL. Control of goal-directed and stimulus-driven attention in the brain. Nature Reviews Neuroscience. 2002;3(3):201–215. doi:10.1038/nrn755

61. Cohen JR, Gallen CL, Jacobs EG, Lee TG, D’Esposito M. Quantifying the reconfiguration of intrinsic networks during working memory. PLoS One. 2014;9(9):e106636. doi:10.1371/journal.pone.0106636

62. Baddeley A. Recent developments in working memory. Current Opinion in Neurobiology. 1998;8(2):234–238. doi:10.1016/S0959-4388(98)80145-1

63. Cook AJ, Im HY, Giaschi DE. Large-scale functional networks underlying visual attention. Neuroscience and Biobehavioral Reviews. 2025;173:106165. doi:10.1016/j.neubiorev.2025.106165

64. Kim S, Lee D. Prefrontal cortex and impulsive decision making. Biological Psychiatry. 2011;69(12):1140–1146. doi:10.1016/j.biopsych.2010.07.005

65. Yang S, Chang MC. Chronic Pain: structural and Functional Changes in Brain Structures and Associated Negative Affective States. International Journal of Molecular Sciences. 2019;20(13):1 doi:10.3390/ijms20133130.

66. Apkarian VA, Hashmi JA, Baliki MN. Pain and the brain: specificity and plasticity of the brain in clinical chronic pain. Pain. 2011;152(3 Suppl):S49–s64. doi:10.1016/j.pain.2010.11.010

Continue Reading

October 25, 2025

Net Profit Margin Jumps to 20.7%, Challenging Profitability Debates

Enova International (ENVA) delivered a 62.4% gain in earnings for the past year, rebounding from an annual decline of 12% over the previous five years. Net profit margins climbed to 20.7%, improving from last year’s 15.4%, and revenue is projected to surge 39.6% per year, outpacing the broader US market’s 10% forecast. With high-quality earnings, accelerating profits, and earnings projected to rise another 13.9% annually, investors are taking stock of Enova’s momentum. The share price of $124.70 currently trades above the estimated fair value of $71.01.

See our full analysis for Enova International.

Next, we will see how this performance compares with the broader narratives that investors and analysts are discussing. Sometimes the numbers shake things up, and sometimes they settle the debate.

See what the community is saying about Enova International

NYSE:ENVA Revenue & Expenses Breakdown as at Oct 2025

Analysts estimate profit margins will contract from 18.8% now to just 7.5% in three years, even as revenue is expected to grow by 60.7% per year through the same period.

According to the analysts’ consensus view, Enova’s technology-driven risk controls and digital platform have supported high margins so far.

However, they debate whether volume gains can continue to offset anticipated pressures from rising regulatory scrutiny, competitive threats, and changing consumer preferences.

This margin squeeze could test the bullish thesis that the company’s underwriting edge and online-only business model will protect profitability over time.

Relatively high current net profit margins of 20.7% remain above last year’s 15.4%, but analysts expect that industry pressures and evolving regulation could challenge Enova’s ability to sustain these levels moving forward.

The consensus narrative flags Enova’s use of advanced AI and real-time analytics for credit risk, enabling rapid adaptation and supporting lower default rates as a key strategic advantage.

Analysts’ consensus view points to the company’s growing share in small business lending, where segment diversification and digital scaling are delivering record origination and consistent credit performance.

This strengthens the argument that Enova can outpace traditional lenders, especially as more customers prefer the speed and convenience of digital-only offerings.

However, expansion into these segments may also bring increased competition from both banks and fintechs, making ongoing technology investment crucial to protecting margins.

Enova’s 10.6x Price-to-Earnings ratio is well below its peer average of 19.8x, though shares at $124.70 currently trade substantially above DCF fair value of $71.01, exposing a 75% premium to fair value and a 7% discount to the analyst price target of $133.63.

Analysts’ consensus view notes this market premium reflects both recent growth outperformance and optimism that digital efficiency and scaling can drive upside.

Yet the valuation gap to fair value remains a watch item, especially as growth normalizes and the company faces sector headwinds not fully captured in current sentiment.

Bulls may argue that Enova’s faster-than-market growth track and tech edge justify the multiple, while skeptics cite margin forecasts and calls for caution on future returns.

With a share price exceeding calculated intrinsic value but remaining below the analyst target, the next stage of the story turns on whether the company can deliver on both its technology edge and profit forecasts to close the valuation gap. See how the bull and bear cases stack up in the community’s narrative for Enova: 📊 Read the full Enova International Consensus Narrative.

To see how these results tie into long-term growth, risks, and valuation, check out the full range of community narratives for Enova International on Simply Wall St. Add the company to your watchlist or portfolio so you’ll be alerted when the story evolves.

Think the numbers tell another story? Share your perspective and shape an independent narrative in just a few minutes: Do it your way.

A great starting point for your Enova International research is our analysis highlighting 3 key rewards and 2 important warning signs that could impact your investment decision.

Enova’s strong topline growth is clouded by concerns about shrinking profit margins and a share price that currently trades well above fair value.

If valuation risks give you pause, use these 877 undervalued stocks based on cash flows to zero in on stocks that market pricing suggests could offer much better value now.

This article by Simply Wall St is general in nature. We provide commentary based on historical data and analyst forecasts only using an unbiased methodology and our articles are not intended to be financial advice. It does not constitute a recommendation to buy or sell any stock, and does not take account of your objectives, or your financial situation. We aim to bring you long-term focused analysis driven by fundamental data. Note that our analysis may not factor in the latest price-sensitive company announcements or qualitative material. Simply Wall St has no position in any stocks mentioned.

Companies discussed in this article include ENVA.

Have feedback on this article? Concerned about the content? Get in touch with us directly. Alternatively, email editorial-team@simplywallst.com

Continue Reading

October 25, 2025
VC bet on $3 billion AI firm ElevenLabs after one meeting with founder

Carles Reina, GTM manager at Eleven Labs, shared why he invested in AI company.

Eleven Labs

The angel investor who backed a billion-dollar AI startup when it was still in its infancy said he decided to invest in the company after just 30 minutes of meeting one of its founders.

Carles Reina first decided to invest in AI voice startup Eleven Labs in 2022, when he was a venture partner at pre-seed fund Concept Ventures.

Co-founded in 2022 by Mati Staniszewski and Piotr Dąbkowski, Eleven Labs specializes in advanced text-to-speech and voice cloning technology. In its January Series C funding round earlier this year, the company raised $180 million at a valuation of $3.3 billion.

Then in September, the company announced it was letting employees sell shares at a $6.6 billion valuation.

However, before Eleven Labs even had a concrete product, Reina, who was working at Palantir Technologies at the time, decided to take a chance on the firm after meeting Staniszewski.

“I met Mati when he was still at Palantir,” Reina told CNBC Make It in an interview. “We started talking, and within 30 minutes of the first conversation, I told him, ‘How much money do you want?’”

Reina explained that before the launch of ChatGPT, voice AI hadn’t garnered much attention because big tech companies like Google, Amazon, and Microsoft all had text-to-speech products, but they hadn’t really taken off.

“With ElevenLabs no one was looking at voice AI, literally no one wanted to give [them] money. No VCs wanted to actually back ElevenLabs, back in the early days on the pre-seed round. So those are the type of industries that I really like, so that I can get in before everyone else,” he said.

Reina has made 74 angel investments over the past eight years, including Revolut, Volumetric, Elroy Air, and Speckle. He now works for Eleven Labs as a go-to-market manager.

He said he always tries to identify industries that other investors are not paying attention to: “I’ve done [invested in] mostly AI before it was sexy. I’ve done robotics before it was sexy as well.”

The No.1 trait to watch for in founders

Reina specializes in investing in pre-seed companies — those with an idea, but often without a fully developed product. This means identifying key traits in founders that indicate a startup will succeed.

“If there is a product, fantastic, but if there is no product, absolutely fine for me … I love founders that are very technical. They’re super sharp, very smart, literally trying to build a global company from day one,” Reina explained.

He said he “invests based on thesis,” so if a founder is very technical, they’ll have a deeper understanding of the product and the market they’re selling to.

Reina said he saw these traits in Staniszewksi, which convinced him to back ElevenLabs despite the voice AI market being very small at the time.

“No one wants to talk to AI voices if they sound robotic. That’s fundamentally the biggest problem that there was right… so when I spoke with Mati, he talked about both elements, and he had not been in the market,” Reina said.

“It was really interesting to see he was thinking about the problems of the entire ecosystem before even actually having any product or before even actually talking to any real potential customer.”

Staniszewski had a background in mathematics with a first-class honors degree from Imperial College London. His vision and technical expertise sold Reina, and ElevenLabs became one of the few startups that he decided to back “literally within less than an hour.”

Now, Eleven Labs is planning a global expansion, including building new hubs in Paris, Singapore, Brazil and Mexico, as well as getting the company ready for IPO within the next five years, Staniszewski told CNBC in July.

Continue Reading

October 25, 2025