Category: 3. Business

Clinically meaningful improvements in longer-term quality of life were achieved with the addition of atezolizumab immunotherapy to chemoradiation in patients with limited-stage small cell lung cancer (LS-SCLC), according to patient-reported outcome findings from the NRG LU005 trial presented during the 2025 American Society for Radiation Oncology (ASTRO) Annual Meeting (Abstract LBA 08). 

“While this study was not randomized between twice-daily vs once-daily radiation, these quality-of-life findings suggest that, relative to once-daily radiation, twice-daily radiation is associated with quality-of-life advantages from the patient perspective,” stated study Quality-of-Life Chair Benjamin Movsas, MD, Medical Director of Henry Ford Cancer, and Chair of Radiation Oncology at Henry Ford Health in Detroit.

Background and Study Analyses Methods

The phase III NRG LU005 trial enrolled 544 patients who were randomly assigned to receive standard chemoradiotherapy of platinum/etoposide every 3 weeks for four cycles plus thoracic radiotherapy at 45 Gy twice daily or 66 Gy daily starting with the second cycle of chemotherapy with or without atezolizumab every 3 weeks for a year starting with the second cycle of chemotherapy. 

Results of the primary endpoint analysis, which were presented at the 2024 ASTRO Annual Meeting, showed that the addition of atezolizumab did not lead to an improvement in overall survival over chemoradiotherapy alone in patients with limited-stage small cell lung cancer. However, an exploratory analysis from the study did show a survival benefit for twice-daily vs daily radiotherapy of 35.4 months vs 28.3 months, respectively. But the difference in received radiation doses was not subject to randomization. 

Patient-reported outcome analyses were planned to explore how the regimens impacted quality of life. Patient-reported outcome tools included FACT-TOI, EQ-5D-5L for quality-adjusted survival, and PROMIS-Fatigue. These measurements were administered at baseline; after chemoradiotherapy; and at 3, 6, 16, and 21 months after chemoradiotherapy. Clinically meaningful declines were defined as a 5-point decline from baseline in FACT-TOI. 

PRO Findings 

At baseline, compliance with patient-reported outcome questionnaires was over 85% and was 60% to 68% through 21 months after chemoradiotherapy. Higher completion rates correlated with better baseline performance status and pulmonary function. 

During chemotherapy, declines in FACT-TOI were observed in both arms. However, they improved by 3 months after treatment and remained stable or improved from baseline levels by 6 to 21 months after treatment. 

At 21 months, fewer patients in the added atezolizumab arm had clinically meaningful declines in FACT-TOI than in the standard chemoradiotherapy alone arm (25% vs 38%). Quality-adjusted survival measurements were similar in both treatment arms. PROMIS-Fatigue showed that immunotherapy did not increase fatigue levels. 

Twice-daily radiotherapy, which was received by about 50% of all patients, was associated with better quality-of-life metrics than daily radiotherapy at all timepoints assessed. Clinically meaningful declines were significantly lower with twice-daily radiotherapy at the end of chemoradiotherapy (36% vs 60%), at 15 months (28% vs 41%), and at 21 months (22% vs 39%). 

Multivariable assessment showed that twice-daily radiotherapy, cisplatin use, and immunotherapy were all significant predictors of lower, clinically meaningful declines. 

Disclosure: The study was supported by the National Cancer Institute of the National Institutes of Health. For full disclosures of the study authors, visit amportal.astro.org. 

UAE, Dubai – 6 October 2025: The Ministry of Economy and Tourism, the New Economy Academy and Ignyte – the region’s leading global digital start-up and SME ecosystem platform developed by Dubai International Financial Centre (DIFC) – announced the launch of the Entrepreneurship Programme, targeting Emirati entrepreneurs.
 
The programme is part of ‘The Emirates: The Start-up Capital of the World’ national campaign, launched by His Highness Sheikh Mohammed bin Rashid Al Maktoum, Vice President and Prime Minister of the UAE and Ruler of Dubai, to strengthen the UAE’s position as a leading global hub for entrepreneurship. The campaign encompasses a comprehensive set of programmes and initiatives designed to empower Emirati youth to launch their ventures, drive innovation and further diversify the UAE economy.
 
The launch and details of the programme were announced at a press conference held in Dubai today, in the presence of His Excellency Abdulla Ahmed Al Saleh, Undersecretary of the Ministry of Economy and Tourism; His Excellency Arif Amiri, CEO of Dubai International Financial Centre Authority and Dr. Laila Faridoon, CEO of the New Economy Academy.

Comprising of a three-day ‘Essentials of Entrepreneurship’ programme and a six-day ‘Complete Entrepreneurship’ masterclass, the Entrepreneurship Programme aims to train 10,000 Emiratis. This initiative combines entrepreneurship fundamentals and practical applications for aspiring entrepreneurs, alongside an advanced track covering the full journey from start-up creation to sustainable growth and global expansion.
 
Participants will hone their entrepreneurship skills, empowering them to launch high-impact projects that can grow locally and compete globally. Ultimately, these efforts will help provide thousands of economic opportunities and accelerate the contributions of SMEs to non-oil GDP in line with the UAE’s aims of bolstering a knowledge-based economy and position itself as a global hub for entrepreneurship and innovation.
 
Innovative ideas
His Excellency Abdulla Ahmed Al Saleh, Undersecretary of the Ministry of Economy and Tourism stated that the Entrepreneurship Programme solidifies the vision of the UAE leadership to establish the country as the world’s capital of entrepreneurship and innovation.
 
His Excellency Al Saleh said: “We believe that investing in our youth and empowering them with modern knowledge tools is the cornerstone for creating a sustainable and competitive creative economy. We are working with our public and private sector partners to provide an integrated ecosystem that enables Emirati entrepreneurs to launch innovative ideas capable of competing regionally and globally, further reinforcing the UAE’s standing as a premier global hub for entrepreneurship.”
 
He added: “Training 10,000 Emiratis in entrepreneurial skills is a strategic move to drive our national talent’s contribution to economic development. This initiative empowers them to launch high-impact ventures that contribute to diversifying our national economy and unlock new avenues for growth.”
 
Exceptional opportunity for Emirati youth
His Excellency Arif Amiri, Chief Executive Officer of Dubai International Financial Centre Authority, emphasised that the Entrepreneurship Programme offers an exceptional opportunity for Emirati youth to foster successful enterprises.
 
His Excellency Amiri said: “The ‘Entrepreneurship Programme’ represents a unique opportunity for Emirati youth to acquire the practical skills and technical knowledge that enable them to transform their ideas into innovative, scalable ventures. This initiative highlights the growing confidence in the capabilities of Emiratis, while enhancing the UAE’s position as a global capital for entrepreneurship and innovation.

As DIFC continues to solidify Dubai’s standing among the world’s top four cities in fintech and innovation, we are proud to be a key partner in this national programme through the Ignyte platform. Attracting 10,000 participants to this platform underscores the nation’s commitment to building a new generation of entrepreneurs capable of spearheading transformation within the economy, in line with the Dubai Economic Agenda (D33) objectives.”

Transforming ideas into projects
Dr. Laila Faridoon, CEO of the New Economy Academy, said: “Through the Entrepreneurship Programme, we seek to equip Emirati youth with the tools and skills needed to transform their ideas into projects on the ground. These efforts further cement the UAE’s position as the destination of choice for entrepreneurs and investors worldwide.”
 
Dr. Faridoon added: “This is the first step towards becoming an entrepreneur. Participants will gain a deeper understanding of innovation methodologies, building successful business models and navigating potential challenges. The programme aims to empower them to become a key part of the UAE’s development journey, and create added value for the national labour market.”
 
Entrepreneurship Programme –Essentials
The Essentials of Entrepreneurship programme equips participants with the fundamental skills and knowledge needed to begin their entrepreneurial journey, professionally and confidently.
 
Held remotely on 19 and 20 October, and in-person at the New Economy Academy at Emirates Towers, Dubai, on 21 October, the programme covers the practical aspects of entrepreneurship, from the early stages of idea generation, business model development, and pitching projects to a panel.
 
The programme targets young innovators, start-ups and existing entrepreneurs seeking to hone their skills and expand their reach.
 
Through workshops and both theoretical and practical training, the programme addresses the fundamentals of entrepreneurship, its regulations and laws in the UAE, market analysis, financial planning, business model development and presentation and marketing skills. The training will be delivered by top entrepreneurship experts in the UAE.
 
Participants will be able to create realistic business plans, as well as financial and marketing plans, and will be able to test their ideas and receive consumers’ feedback. Graduates will receive a certificate from the New Economy Academy.
 
Entrepreneurship Programme – Complete
The Complete Entrepreneurship masterclass, among the largest of its kind in the region, is a comprehensive training experience that covers all stages of building a business, from idea generation to sustainable growth and expansion. The 6-day course combines in-person and remote sessions.

In-person sessions will take place from 21 to 23 October, at the New Economy Academy at the Emirates Towers in Dubai, while remote sessions will run from 28 to 30 October.
This programme is ideal for aspiring entrepreneurs seeking to build locally and globally scalable ventures.
 
Combining theory with hands-on practice, the programme addresses all stages of entrepreneurship, from identifying business ideas, to market and competitor analysis, marketing strategies and finally pitching projects to expert panels.
 
Main themes of the programme include identifying opportunities and assessing risks, building revenue models, idea and market validation, company formation laws, Go-to-Market strategies and integrated marketing campaigns.
 
‘The Emirates: The Start-up Capital of the World’ national campaign is supervised by the Ministry of Economy and Tourism in collaboration with the UAE Government Media Office, with the participation of the UAE Council for Entrepreneurship, public and private sector partners and national organisations.

Developed by DIFC, Ignyte is set to become a cornerstone of the Dubai Digital Economy Strategy, empowering start-ups and entrepreneurs on their journey toward global growth, with the vision to support over 100,000 start-ups by connecting them with over 5,000 mentors and investors by 2029.
 
It offers entrepreneurs invaluable guidance, and enables start-ups to connect with potential investors, global mentors, participate in networking events, and tap into exclusive offers that provide significant cost savings.

 

The FDA has accepted and granted priority review to a biologics license application (BLA) seeking the approval of the allogeneic T-cell immunotherapy Orca-T for the treatment of select patients with hematologic malignancies, including acute myeloid leukemia (AML), acute lymphoblastic leukemia (ALL), and myelodysplastic syndromes (MDS).¹

The BLA is supported by data from the phase 3 Precision-T trial (NCT04013685), which demonstrated that patients with AML, ALL, and MDS treated with Orca-T achieved a statistically significant improvement in moderate-to-severe chronic graft-vs-host disease (cGVHD)–free survival compared with those who received conventional allogeneic hematopoietic stem cell transplant (allo-HSCT).

Topline data from Precision-T showed that the 1-year moderate-to-severe cGVHD-free survival rate was 78% (95% CI, 65%-87%) in the Orca-T arm (n = 93) vs 38% (95% CI, 26%-51%) in the allo-HSCT arm (HR, 0.26; P < .00001).² The cumulative incidence of moderate-to-severe cGVHD was 13% (95% CI, 5%-23%) with Orca-T vs 44% (95% CI, 31%-56%) with allo-HSCT.

Regarding overall survival (OS)—a secondary end point—the 1-year rates were 94% (95% CI, 86%-97%) and 83% (95% CI, 73%-90%) in the Orca-T and allo-HSCT arms, respectively (HR, 0.49; P = .11823).

The FDA has assigned a target action date for the BLA of April 6, 2026, under the Prescription Drug User Fee Act.¹

“[Allo-HSCT] has been the only potentially curative option for many people with AML, ALL or MDS, however treatment-related toxicities too often hinder patient recovery. Acceptance of the Orca-T BLA marks a pivotal moment in our ability to deliver a first-in-class therapy designed to improve survival free from complications like GVHD,” Nate Fernhoff, PhD, cofounder and chief executive officer at Orca Bio, stated in a news release. “Supported by positive phase 3 clinical data, today’s regulatory milestone reflects important recognition of the transformative potential of Orca-T. We look forward to working collaboratively with the FDA on the review of our application with the goal of advancing Orca-T and making it available to patients in need.”

How Was the Precision-T Trial Designed?

The multicenter, randomized, open-label Precision-T study enrolled patients 18 to 65 years of age with acute leukemia in complete remission (CR) or CR with incomplete hematologic recovery; or with MDS that is indicated for allo-HSCT per 2017 International Expert Panel recommendations and/or therapy-related/secondary MDS, with no more than 10% bone marrow blasts.³ Patients needed to be planning to undergo a matched related or unrelated donor allo-HSCT with total body irradiation (TBI) and cyclophosphamide; TBI and etoposide; or busulfan, fludarabine, and thiotepa.

Key inclusion criteria comprised a resting cardiac ejection fraction of at least 45% or a shortening fraction of at least 27%; alanine aminotransferase and aspartate aminotransferase levels less than 3 times the upper level of normal (ULN); intermediate- or high-risk disease designation; and a total bilirubin level less than the ULN.

Investigators excluded patients who received a prior allo-HSCT, those with a planned donor lymphocyte infusion, and those with planned pharmaceutical in vivo or ex vivo T-cell depletion.

Patients were randomly assigned to receive Orca-T plus single-agent tacrolimus; or alloHSCT plus tacrolimus and methotrexate.² In both arms, patients underwent myeloablative conditioning and used a related or unrelated matched donor.

Along with the primary end point of moderate-to-severe cGVHD-free survival at 1 year, secondary end points comprised time to moderate-to-severe cGVHD; OS; and the rate of patients free from both cGVHD and relapse at 1 year.³

What Other Data Have Been Reported From Precision-T?

Topline data also showed that the 1-year relapse-free survival rate was 76% in the Orca-T arm compared with 74% in the alloHSCT arm (HR, 0.80; P = .49).² The cumulative rate of non-relapse mortality was 3% for patients treated with Orca-T vs 13% for those who underwent allo-HSCT. The rates of grade 3/4 acute GVHD were 6% and 17% in the Orca-T and allo-HSCT arms, respectively.

Regarding safety, no new safety issues were reported for Orca-T. Grade 4 or higher infections occurred in 6% of patients in the Orca-T arm vs 10% of patients in the allo-HSCT arm.

References

1. Orca Bio announces FDA acceptance and priority review of the biologics license application (BLA) for Orca-T to treat hematological malignancies. News release. Orca Bio. October 6, 2025. Accessed October 6, 2025. https://orcabio.com/orca-bio-announces-fda-acceptance-and-priority-review-of-the-biologics-license-application-bla-for-orca-t-to-treat-hematological-malignancies/
2. Orca Bio announces positive results from the pivotal phase 3 study of investigational Orca-T compared to allogeneic stem cell transplant for the treatment of hematologic malignancies. News release. Orca Bio. March 17, 2025. Accessed October 6, 2025. https://orcabio.com/orca-bio-announces-positive-results-from-the-pivotal-phase-3-study-of-investigational-orca-t-compared-to-allogeneic-stem-cell-transplant-for-the-treatment-of-hematologic-malignancies/
3. Precision-T: a randomized study of Orca-T in recipients undergoing allogeneic transplantation for hematologic malignancies (Orca-T). ClinicalTrials.gov. Updated September 26, 2025. Accessed October 6, 2025. https://clinicaltrials.gov/study/NCT05316701

Chemistry

The compounds were synthesized following the reported procedures [16]. (Z)−4-arylidene-2-substituted oxazol-5(4 H)-one intermediates (3) were prepared by Erlenmeyer method through reaction of N-acetyl glycine (1) or hippuric acid with the appropriate aldehyde (2) in acetic anhydride and sodium acetate. Kim Compounds (5a-d) were prepared by cyclocondensation reaction of (Z)−4-arylidene-2-substituted oxazol-5(4 H)-one intermediates (3) with 4-amino-N-benzylphenyl acetamide (4) in dry pyridine to yield the target molecules (5a-d) [16]. These four compounds are Z type and include 5a (KIM-161): (Z)-N-Benzyl-2-(4-(4-(4-methoxybenzylidene)−2-methyl-5-oxo-4,5-dihydro-1 H-imidazol-1-yl)phenyl)acetamide, 5b (KIM-111): (Z)-N-Benzyl-2-(4-(4-benzylidene-2-methyl-5-oxo-4,5-dihydro-1 H-imidazol-1-yl)phenyl)acetamide, 5c(KIM-261): (Z)-N-Benzyl-2-(4-(4-(4-methoxybenzylidene)−5-oxo-2-phenyl-4,5-dihydro-1 H-imidazol-1-yl)phenyl)acetamide, 5 d (KIM-231): (Z)-N-Benzyl-2-(4-(4-(4-hydroxy-3-methoxybenzylidene)−5-oxo-2-phenyl-4,5-dihydro-1 H-imidazol-1-yl)phenyl)acetamide.

All the target compounds were confirmed by different analytical techniques including spectral analysis using nuclear magnetic resonance (NMR), and the molecular formulae were detected using high-resolution mass spectrometry (HRMS). The purity of all compounds was confirmed by liquid chromatography/mass spectrometry (LC/MS) to be higher than 95%. Scheme 1 shows the synthetic method of our target compounds (5a-d) and their project codes. The synthetic procedures and detailed characterization of all compounds have been comprehensively described in our earlier work [16].

Physicochemical properties

The physicochemical properties explained the molecular weights, molecular formula, number of heavy atoms, number of aromatic heavy atoms, fraction Csp3, number of rotatable bonds, number of hydrogen bonds, molar refractivity, topological polar surface area, and other physicochemical features shown in Table 1. They also displayed the high lipophilic characters of these compounds which are responsible for high GIT absorption and BBB permeation especially in the two compounds Kim-161, Kim-111. They also have moderate water solubility, good bioavailability score, moderate synthetic accessibility, and obey Lipiniski rule which means these compounds are suitable for production as drug. On the other side, the two compounds Kim-231, Kim-261 have poor water solubility, good bioavailability score, moderate synthetic accessibility, and have two violations from Lipiniski rule which means these compounds have no optimum characters for production as drug. Table 1 shows the physicochemical properties of the target compounds calculated by using SwissADME [29].

TPSA = topological polar surface area. Bioavailability image: The colored zone is the suitable physiochemical space for oral bioavailability: Lipo = lipophilicity, polar = polarity, insole = insolubility, insatu = instauration, Flex = flexibility.

Biological screening

MTT assay

A comparative investigation of proliferation inhibition across cancer cell line T24 by the MTT assay [30] for Kim-111, Kim-161, Kim-231, and Kim-261 molecules revealed that both Kim-111 and Kim-161 drugs exhibited cytotoxicity to the cells, demonstrating varying inhibition percentages with IC₅₀ = 67.29 µM for Kim-111, and IC₅₀ = 56.11 µM for Kim-161. The inhibitory impact considerably increased along with the increased concentrations in contrast to Kim-231, and Kim-261which exhibited no inhibitory effect at the concentrations tested in the current set up as illustrated in Fig. 3.

Effect of Kim 111 and Kim 161 compounds on cell senescence and apoptosis in T24 urinary bladder cancer cells

A panel of suggested anticancer pathways were tested for evaluating the impact of Kim 111 and Kim 161 treatment on the genes expression responsible for cell senescence, apoptosis, inflammation and metastasis (Table 2).

The markers of responsible for cell senescence; (p53), oncogenesis (Kras) and cell apoptosis; (BAX and caspase 3) were evaluated to detect the difference between the Kim-treated groups and the control non-treated T24 bladder cancer cells.

P53 gene responsible for cell senescence, Kim 111 treatment showed statistically lower expression in comparison to the untreated T24 cells, but in contrast, the cells treated with Kim 161 showed a dramatic higher expression in comparison to the untreated cells. On a same pattern, treatment with Kim 111 reduced the expression of Kras oncogen significantly but Kim 161 showed statistically higher expression in comparison to the untreated cells. To evaluate the effect of the tested compounds on cell apoptosis, the genes suggested to regulate the apoptosis process were investigated following the T24 cells treatment, namely BAX and caspase 3. The data demonstrated that cells treated with Kim 111 and 161 demonstrated a higher expression with statistically significant difference between both treated groups, and in comparison, to the untreated cells (Fig. 4).

Effect of Kim 111 and Kim 161 compounds on the inflammatory markers in T24 urinary bladder cancer cells

Regarding the inflammatory genes, IL6 showed statistically lower expression in the cells treated with Kim 111 in comparison to the untreated cells, but in contrast, the cells treated with Kim 161 showed statistically higher expression in comparison to the untreated cells. TNFα and NF-κB1 genes expression increased following the treatment with Kim 111 and Kim116, with statistically significant difference between both treated groups, and in comparison, to the untreated cells (Fig. 5).

Effect of Kim 111 and Kim 161 compounds on the autophagy and metastasis markers in T24 urinary bladder cancer cells

The current study assessed the role of Kim 111 and Kim 161 in regulating cell proliferation and survival through evaluating the phosphoinositide 3-kinase (PIK3CA) and its target genes; Akt, and mTOR. Both the Kim 111 and Kim 161- treated groups showed significantly lower PIK3CA and mTOR expression relative to the untreated group. There was no statistically-significant difference in the expression of the PIK3CA and mTOR genes between the cells treated with Kim 111 and those treated with Kim 161. In contrast, Akt gene expression showed higher expression with statistically significant difference between both treated groups, and in comparison, to the untreated cells. Regarding the metastatic gene Matrix metalloproteinase-9 (MMP-9), both treated groups showed statistically lower expression in comparison to the untreated cells, with a non-significant difference between cells treated with Kim 111 and Kim 161 (Fig. 6).

Data presented as mean ± standard error of gene expression fold changes of cells′ triplicates. Capital letters indicate p values ≤ 0.05 using One-Way ANOVA with Tukey post-hoc testing (similar letters indicate non-significant difference, different letters denote significant difference).

The above-mentioned data demonstrated the potential of novel imidazole derivatives, Kim-161 and Kim-111, as promising anticancer agents against urothelial carcinoma (T24 cells). Clinicians may consider Kim-161 over Venetoclax if ongoing trials confirm enhanced BCL-2 inhibition, improved safety, or efficacy in resistant cancers. We hope its optimized structure offers better pharmacokinetics, reduced toxicity, or broader antitumor activity, addressing Venetoclax limitations. However, further studies are needed to validate these potential advantages.

Computational modeling

Docking and binding poses

The top three targets identified (PTK6, FLT3, and BCL-2) correspond to the crystal structures 6CZ4, 4XUF, and 6O0K, respectively. The targets PTK6 (6CZ4), FLT3 (4XUF), and BCL-2 (6O0K) were selected due to their critical roles in urothelial carcinoma (UC) progression. PTK6 promotes cell proliferation and survival, making it a key oncogenic kinase. FLT3, primarily associated with leukemia, is implicated in UC through aberrant signaling pathways. BCL-2, an anti-apoptotic protein, is often overexpressed in UC, contributing to chemoresistance. The crystal structures 6CZ4, 4XUF, and 6O0K provide high-resolution templates for docking studies, enabling the evaluation of novel derivatives as potential inhibitors. Targeting these proteins may disrupt proliferation and survival mechanisms, offering a strategic approach for UC therapy. Glide XP docking scores for Kim-111 and Kim-161 in these sites are summarized in Table 3 alongside the scores of the native co-crystallized ligands. Overall, Kim-111 showed very favorable docking to the two kinases PTK6 and FLT3 (XP GScore ≈ − 11.3 and − 11.6 kcal/mol, respectively), essentially matching the binding score of the native inhibitors (–14.76 for PTK6’s ligand FKY, − 11.75 for FLT3’s Quizartinib). Kim-161 likewise docked well in those sites (≈ − 11.6 kcal/mol). In BCL-2, the docking scores for Kim-111 and Kim-161 were more modest (around − 7.4 to − 7.9 kcal/mol), reflecting weaker predicted binding than the native BCL-2 inhibitor Venetoclax (docking score − 10.94 kcal/mol). Notably, the poses of Kim-161 in BCL-2 overlapped substantially with Venetoclax’s position, indicating Kim-161 can fit in the same hydrophobic pocket that accommodates the BH3-domain of pro-apoptotic proteins. Figure 7 shows the 3D interaction diagram of Venetoclax bound in the BCL-2 (6O0K) pocket after 100 ns MD, however Fig. 8 shows the 3D interaction diagram of Kim-161 bound in the BCL-2 (6O0K) pocket after 100 ns MD. Key interacting residues are highlighted with their interaction frequency over the trajectory. Kim-161 forms strong hydrogen bonds (purple dashed arrows) with Tyr108 (83% occupancy) and Asn143 (66%), mimicking the interactions of the native inhibitor. Hydrophobic contacts (green lines) with Phe104 (57%) and Met115 stabilize the ligand in the hydrophobic groove. These interactions help anchor Kim-161 similarly to Venetoclax, supporting a BCL-2 inhibitory mechanism.

The induced fit docking (IFD) results showed further optimization of these poses. Allowing side-chain flexibility led to small improvements in docking scores for the Kim compounds (Table 3). For instance, in FLT3 the IFD-refined pose of Kim-111 achieved a docking score of − 14.29 kcal/mol (improved from − 11.58), surpassing the native ligand’s re-dock score. Such improvements suggest Kim-111 can induce slight conformational adjustments in the FLT3 binding site to enhance complementarity. In BCL-2, IFD refinement of Kim-161 allowed Tyr108 and Arg146 side chains to shift and form the above-mentioned hydrogen bonds, yielding a more favorable score (–8.84 vs. initial − 7.89). Overall, IFD confirmed that Kim-111 and Kim-161 are capable of binding in the active sites of these targets with only minor local receptor adjustments, and no major backbone motion was required to accommodate them.

Molecular Mechanics Generalized Born Surface Area (MM-GBSA) was used to estimate the binding free energy because it is considered an ideal tool for high-throughput screening to determine structural stability, and predict binding affinities. Being a faster and less expensive computational tool makes it a promising alternative to the more complex free energy calculations and empirical scoring functions.

Table 3 displays the docking scores (Glide XP) and (MM-GBSA) binding energies for Kim-161 and Kim-111 in top targets, compared to native ligands.

Docking scores (in kcal/mol) are from Glide XP (negative values indicate better predicted affinity). ΔG_bind is the binding free energy from Prime MM-GBSA (more negative = more favorable). PTK6 native Ligand refers to PDB 6CZ4’s inhibitor (FKY); FLT3 native is Quizartinib in 4XUF; BCL-2 native is Venetoclax in 6O0K. Figure 9 shows the 3D interaction diagram of Kim-111 bound in the BCL-2 (6O0K) pocket after 100 ns MD.

As shown in Table 3, the MM-GBSA calculations followed the trend of the docking scores. Kim-111 and Kim-161 were predicted to bind most strongly to FLT3 and PTK6 (ΔG_bind ~ − 69 to − 75 kcal/mol), whereas their binding to BCL-2 was relatively weaker (ΔG_bind ~ − 66 kcal/mol). Importantly, all ΔG_bind values are significantly negative, suggesting that both compounds can form stable complexes with each target in an aqueous environment. The native ligands, being high-affinity inhibitors, have substantially more favorable ΔG_bind (e.g., − 122.5 kcal/mol for Venetoclax with BCL-2), which is expected given they were optimized for those targets. Figure 10 illustrates the comparative binding free energies for the two compounds vs. native ligands, highlighting that Kim-111 and Kim-161 achieve strong binding in the kinases and moderately strong binding in BCL-2.

Molecular dynamics stability

The 100 ns MD simulations provided dynamic validation of the docking poses. In all cases, the protein–ligand complexes remained intact over the simulation, with no unbinding events observed. The RMSD profiles (protein Cα and ligand) are shown in Fig. 11 for a representative complex, and quantitative stability metrics from all simulations are summarized in Table 3. Overall, the FLT3 complexes were the most stable: with Kim-111 bound to FLT3, the protein backbone RMSD plateaued around ~ 1.8 Å and the Ligand RMSD stayed below 1.5 Å for nearly the entire 100 ns (Fig. 11). Kim-161 in FLT3 showed a similarly low ligand RMSD (~ 1.0 Å average), indicating that both compounds snugly occupy the FLT3 kinase site without significant displacement. The PTK6 simulations showed a slightly higher backbone RMSD (~ 2.5 Å) but still reasonable stability. Notably, Kim-111 in PTK6 maintained a low ligand RMSD (~ 2 Å), whereas Kim-161 in PTK6 had larger fluctuations (ligand RMSD rising to ~ 4 Å transiently), suggesting Kim-161’s fit in PTK6 is less optimal or stable than Kim-111’s.

In the BCL-2 simulations, the behavior of the two compounds diverged. Kim-161 formed a very stable complex with BCL-2: after an initial adjustment (ligand RMSD < 2 Å in the first 5 ns), Kim-161 remained close to its docked position (average ligand RMSD ~ 1.8 Å) for the rest of the trajectory. In contrast, Kim-111 in BCL-2 exhibited more movement – its Ligand RMSD fluctuated between 2 and 4 Å, albeit remaining in the pocket. This aligns with the weaker docking score and MM-GBSA energy of Kim-111 for BCL-2, suggesting Kim-111 does not engage BCL-2 as firmly as Kim-161 does. The BCL-2 protein itself remained stable (backbone RMSD ~ 1.5 Å) when either ligand was bound, comparable to the control simulation with Venetoclax (which showed backbone RMSD ~ 1.3 Å and ligand RMSD ~ 1 Å, as expected for a very tight binder).

The MD analyses also examined residual flexibility and interactions. RMSF plots (see Supplementary Figures S1) showed that loop regions at the periphery of each binding site had mild fluctuations (up to 3 Å), but critical binding site residues remained relatively rigid (RMSF < 1 Å) in the presence of each ligand. In FLT3, residues in the activation loop and the hinge (which interact with inhibitors) were stabilized by both Kim-111 and Kim-161, like the native inhibitor. In BCL-2, the presence of Kim-161 kept the binding groove residues (e.g., Tyr108, Phe104, Arg146) ordered, whereas with Kim-111 there was slightly more fluctuation in the loop containing Arg146, correlating with Kim-111’s less consistent interactions there.

Crucially, the MD trajectories confirmed that key protein–ligand interactions identified in docking persisted over time. For example, in the BCL-2–Kim-161 simulation, the hydrogen bond between Kim-161’s amide carbonyl and the side chain of Asn143 was maintained ~ 66% of the time, and a stable hydrogen bond (83% occupancy) formed between Kim-161’s secondary amine and the backbone carbonyl of Tyr108. These interactions closely mirror those formed by Venetoclax in BCL-2, lending confidence that Kim-161 indeed engages the BCL-2 pocket in a functionally relevant manner. Kim-111 also retained a hydrogen bond with Tyr108 but with lower frequency (~ 40% of frames), explaining its weaker binding. In FLT3, Kim-111 consistently formed two hydrogen bonds with the kinase hinge region (to a backbone carbonyl and a side-chain of a hinge residue, analogous to Quizartinib’s interactions), which remained > 70% occupied during MD. Kim-161 in FLT3 also maintained hinge region hydrogen bonds, though one interaction was intermittent (about 30% occupancy), consistent with its slightly lower affinity prediction. Hydrophobic contacts were largely preserved, e.g., Kim-111 stayed nestled against FLT3’s gatekeeper residue and Phe830 (in the activation loop) through π-stacking and van der Waals contacts, and Kim-161 remained in contact with BCL-2’s hydrophobic groove residues Phe104 and Val130 over > 80% of the simulation frames. Figures 12 and 13 display the 2D interaction diagram of Kim-161, Kim-111 bound in the FLT-3 (4XUF) pocket after 100 ns MD and molecular dynamic analysis graphs.

In summary, the MD results support the docking findings by demonstrating that Kim-111 and Kim-161 form stable complexes with all three putative targets in an explicit solvent environment. The stability ranking from MD (most stable in FLT3, followed by PTK6, then BCL-2 for Kim-111; and FLT3 ≈ BCL-2 > PTK6 for Kim-161) aligns with the binding affinity predictions. Minor differences in stability (e.g., Kim-161 outperforming Kim-111 in BCL-2, and Kim-111 slightly more stable in PTK6) were observed, highlighting the complementary nature of these two compounds in targeting different proteins. The persistence of critical hydrogen bonds and the low ligand RMSDs especially underscore that Kim-111 is well-suited for targeting FLT3/PTK6, whereas Kim-161 is particularly effective at engaging BCL-2.

Computational modeling provides a mechanistic rationale for the biological activities of Kim-111 and Kim-161. Docking and MD simulations identified multiple potential targets for these compounds, which suggests a polypharmacological mode of action (i.e. hitting more than one target) [31]. Two oncogenic kinases (PTK6 and FLT3) and the anti-apoptotic protein BCL-2 emerged as top binding targets. This aligns well with the experimental observations that Kim-111 and Kim-161 exhibit potent anti-cancer effects in bladder cancer cells, likely through a combination of pro-apoptotic and anti-proliferative mechanisms.

BCL-2 as an apoptosis-related target: The strong binding of Kim-161 to BCL-2 is especially notable. BCL-2 is a key regulator of cell death; inhibition of BCL-2 frees pro-apoptotic factors like BAX/BAK to trigger mitochondrial apoptosis. Our simulations showed that Kim-161 occupies the same site as the known BCL-2 inhibitor Venetoclax, even forming analogous interactions (Tyr108 and Asn143 hydrogen bonds, hydrophobic contact with Phe104). Although Kim-161’s computed affinity is lower than Venetoclax, it is still substantial, and the MD data confirmed the complex remains stable. This provides a molecular basis for the apoptosis induction observed in treated cells – if Kim-161 binds BCL-2, it would neutralize BCL-2’s anti-apoptotic function, leading to increased apoptosis. Indeed, in our biological assays, both compounds increased the expression of pro-apoptotic genes (BAX, Caspase 3) and markers of cell death in bladder cancer cells, consistent with BCL-2 inhibition relieving apoptosis suppression. Interestingly, Kim-161 had a greater effect on these apoptosis markers than Kim-111 (e.g., higher BAX, caspase-3, and p53 expression in Kim-161–treated cells, relative to Kim-111), which correlates with Kim-161’s stronger BCL-2 engagement in silico. We therefore propose that Kim-161 acts as a BH3-mimetic agent targeting BCL-2, triggering intrinsic apoptotic pathways in the cancer cells.

Kinase inhibition and anti-proliferative effects: Both Kim-111 and Kim-161 showed high docking affinity to PTK6 (protein tyrosine kinase 6) and FLT3 (Fms-like tyrosine kinase 3). These kinases are implicated in tumor cell proliferation and survival signaling [30,31,32]. PTK6, for instance, is overexpressed in some cancers and can promote migration and growth, while FLT3 activates pathways like PI3K/Akt and Ras/MAPK in leukemia and potentially other contexts [33,34,35]. Our finding that Kim-111 binds very stably to FLT3 (with low RMSD and persistent hydrogen bonds in the ATP-binding site) suggests Kim-111 could be an effective FLT3 inhibitor. Kim-161 also bound FLT3 well, albeit slightly less tightly. In the biological data, we observed downregulation of PI3K (PIK3CA gene) and mTOR expression in cells treated with either compound, along with changes in Akt and KRAS expression. These could be downstream consequences of kinase inhibition: if Kim-111 and Kim-161 inhibit FLT3 or PTK6, they would dampen PI3K/Akt and RAS signaling, resulting in reduced proliferation and possibly induction of senescence. Notably, Kim-111-treated cells had lower p53 and KRAS expression than controls (indicating a senescence-like cell cycle arrest), whereas Kim-161-treated cells showed an opposite trend for those genes. This divergence might be explained by their target spectrum Kim-111, with stronger kinase (FLT3/PTK6) inhibition, may enforce cell-cycle arrest without heavily stressing apoptotic pathways (hence lower p53 in a feedback loop), while Kim-161, by strongly hitting BCL-2, pushes cells into apoptosis (which can elevate p53 as a DNA damage response). In essence, Kim-111 might act more as a multi-kinase inhibitor, slowing proliferation, whereas Kim-161 has a dual action: moderate kinase inhibition plus potent apoptosis induction via BCL-2. Such multi-targeted kinase inhibitors are common in oncology (e.g., many TKIs hit multiple kinases) [16], and can be advantageous by blocking redundant survival pathways.

Target selectivity and synergy: The differential target binding also provides insight into how Kim-111 and Kim-161 could be used therapeutically. Kim-111’s high affinity for FLT3 suggests potential in malignancies where FLT3 is important (e.g. leukemia), or in solid tumors if FLT3 or PTK6 contribute to growth. Meanwhile, Kim-161’s targeting of BCL-2 indicates it could sensitize cancer cells to apoptosis or even be combined with other therapies. In bladder cancer, BCL-2 is one mechanism of chemo-resistance; thus Kim-161 might enhance the efficacy of chemotherapy by disabling tumor cell survival programs. The fact that both compounds can engage multiple targets (kinase and BCL-2) is promising, as cancer cells often rely on network redundancy. By concurrently attacking survival signaling and anti-apoptotic defenses, Kim-111 and Kim-161 may achieve a one-two punch: forcing cell cycle exit and then driving programmed cell death. This hypothesis is supported by the comprehensive changes seen in gene expression (senescence markers, apoptotic markers, and inflammatory pathways all altered). Our computational data back up this breadth of action with concrete binding interactions at the molecular level. Figures 14, and 15 show the 2D interaction diagram of Kim-161 and Kim-111 bound in the FLT-3 (4XUF) pocket after 100 ns MD and molecular dynamic analysis graphs.

Given that both Kim-111 and Kim-161 are designed as substituted imidazole derivatives with activity against numerous kinases, it is crucial to consider their potential off-target effects, which may arise from unintended interactions with non-target kinases or cellular proteins. These effects raise concern about the therapeutic efficacy, safety and specificity of these compounds [36].

Since the kinase inhibitors often share conserved ATP-binding sites, cross reactivity and or/interactions with non-target kinases. For example, KIM-161 has been reported to downregulated several other kinases such as the ERK1/2, GSK-3α/β, HSP27, and JAK/STAT2 signals which may result in unintended immune modulation [16].

Moreover, the imidazole derivatives are known to interact with cytochrome P450 enzymes, potentially affecting the drug pharmacokinetic and metabolism. The hydrogen binding potential of those derivatives is another factor that raises the possibility of binding to ion channels or G-protein coupled receptors (GPCRs), which could manifest as neurological, endocrinal or gastrointestinal disturbances [37].

Thus, future studies should incorporate in-vivo toxicity assays, selectivity tests and wide molecular docking profiling for a broad panel of kinases to predict any systemic risks and mitigate any side effects to ensure the therapeutic efficacy and safety of those compounds.

The in silico findings from this study should also be followed by targeted experimental validation. For instance, pull-down or SPR assays could confirm direct binding of Kim-161 to BCL-2, and kinase inhibition assays (or cellular phospho-substrate readouts) could verify if FLT3 or PTK6 activity is suppressed by Kim-111. Co-crystallization or cryo-EM of Kim-161 with BCL-2 would definitively show if it bound in the Venetoclax pocket as predicted. Likewise, co-crystal structures with FLT3 or PTK6 would guide optimization of the scaffold for even tighter binding as our modeling suggests, for example, that adding functionality to interact with PTK6’s Asp164 (hinge region) might boost Kim-161’s affinity for PTK6. Medicinal chemistry optimization could proceed differently for the two compounds: for Kim-161, increasing affinity to BCL-2 (perhaps by extending into the P2 pocket like Venetoclax’s chlorophenyl group) could yield a potent apoptotic agent, whereas for Kim-111, maintaining broad kinase coverage but improving BCL-2 activity might create a balanced dual-target drug. Given the computational insight that each compound excels on different targets, designing analogs that combine the favorable features of both (kinase inhibition of Kim-111 with the BCL-2 inhibition of Kim-161) is an intriguing strategy.

In conclusion, our computational modeling study provides a detailed picture of how Kim-111 and Kim-161 likely interact with cancer-related proteins at the molecular level. The data supports a mechanism where Kim-161 directly antagonizes BCL-2, unleashing apoptosis, while Kim-111 potently inhibits oncogenic kinases like FLT3/PTK6, impeding survival signaling and both compounds exhibit overlapping activities across these targets. This dual modality is highly consistent with the observed anti-proliferative and pro-apoptotic effects in bladder cancer cells. The most promising target identified is BCL-2, given its pivotal role in apoptosis and the strong evidence of Kim-161 binding; targeting BCL-2 aligns with the pronounced cell death (and upregulation of apoptosis markers) seen experimentally. PTK6 and FLT3 emerge as additional targets that could contribute to the compounds’ efficacy by reducing cell viability and invasiveness (for instance, PTK6 is linked to migration and its inhibition might reduce metastatic potential). By correlating in silico predictions with in vitro observations, we have built a cohesive rationale for the multifaceted anti-cancer action of Kim-111 and Kim-161. These insights not only validate the compounds’ mechanisms but also guide future development suggesting that further target validation (especially BCL-2) and structure-based optimization could eventually lead to new therapeutic candidates for bladder cancer and possibly other cancers. The computational approach applied here, integrating docking, induced fit refinement, MM-GBSA scoring, and long-timescale MD, exemplifies how in silico modeling can profoundly inform and accelerate drug discovery efforts in identifying and characterizing drug-target interactions in the absence of exhaustive experimental structural biology.

This study aimed to compare the expression levels of TROP-2 and Nectin-4 in patients with upper tract urothelial carcinoma (UTUC) who had a prior history of urinary bladder carcinoma (UB-Ca) to those without such a history. While TROP-2 and Nectin-4 are well-studied therapeutic targets in UB-Ca, their comparative expression patterns in UTUC subgroups remain underexplored, particularly in patients with a history of UB-Ca. Our findings addressed this gap and highlighted critical differences with potential clinical implications.

We demonstrated that Nectin-4, the target protein of enfortumab vedotin, was expressed in 70.1% of patients with UTUC. This finding is consistent with the results of Tomiyama et al., who reported an expression rate of 65.7% in their cohort [10]. In UB-Ca, however, Nectin-4 expression has been documented in approximately 83% of cases, suggesting a slightly higher expression rate than UTUC [11]. Intriguingly, UTUC patients with prior UB-Ca history showed a non-significantly higher Nectin-4 expression rate (74.1% vs. 63.6%, p > 0.05). This trend mirrors findings by Klümper et al., who observed reduced Nectin-4 expression in metastatic urothelial carcinoma, potentially linked to EV resistance mechanisms [8]. While statistical significance was not achieved, this trend underscores the need for larger cohort studies to clarify the role of UB-Ca history in UTUC biology.

Powles et al. previously reported a median H-Score of 280 in patients with locally advanced or metastatic urothelial cancer, without distinguishing between UTUC and UB-Ca, in their exploratory analysis of Nectin-4 expression’s impact on outcomes with enfortumab vedotin (EV) plus pembrolizumab (P) in the Phase 3 EV-302 study [9]. Our cohort, however, showed markedly lower H-scores (mean 66), with 78.7% of patients falling into low-expression categories. This discrepancy may reflect both biological and methodological factors. Biologically, UTUC may inherently exhibit lower Nectin-4 expression due to its distinct urothelial differentiation like tumor stage (our cohort included localized UTUC vs. EV-302’s metastatic cases), immune context, or tumor evolution compared to bladder tumors [12,13,14]. Methodologically, variations in tissue fixation, antibody clones, or H-score evaluation across studies may contribute to the observed differences [13]. These findings emphasize the importance of tumor-site- and disease-stage-specific validation of Nectin-4 as a therapeutic biomarker before applying ADC-based treatment strategies. Despite the H-score widespread application, establishing standardized H-score thresholds for clinical responses remains a subject of debate, primarily due to variability in scoring systems across different institutions.

Recent studies have shown that TROP-2, the target protein of sacituzumab govitecan, is widely expressed in UTUC, with 94% of UTUC cases demonstrating positivity. High TROP-2 expression has also been associated with favorable prognosis in UTUC [15]. In our study, TROP-2 expression was detected in 98.8% (86 out of 87 patients), confirming the high expression rate of TROP-2 in UTUC, which is higher than that of Nectin-4. Notably, our findings were consistent with Tomiyama et al., who observed stronger TROP-2 expression (95.6%) in low-grade UTUC compared to high-grade variants, which was associated with a favorable prognosis [15]. This contrasts with findings in other cancers, such as non-muscle-invasive UB-Ca, breast cancer, and metastatic prostate cancer, where high TROP-2 expression has been linked to increased tumor aggressiveness and poor prognosis [15,16,17,18,19,20]. Recent UTUC studies indicate that high TROP-2 expression may instead reflect a more differentiated, luminal-like phenotype associated with favorable outcomes in this tumor type [15]. Differences in subcellular localization (membranous vs. cytoplasmic), signaling partners, and co-expression with luminal markers may modulate its function and prognostic implications in UTUC [15, 20]. In our cohort, no significant associations were found between TROP-2 expression and clinicopathological factors such as lymphovascular invasion or lymph node or distant metastasis.

Additionally, subgroup analysis based on UB-Ca classification did not reveal significant differences in Nectin-4 or TROP-2 expression. Although patients with a positive history of UB-Ca exhibited higher intensity of TROP-2 expression, the difference was not statistically significant. This suggests that a history of UB-Ca does not directly influence the expression levels of TROP-2 or Nectin-4 in UTUC.

The role of prior intravesical Bacillus Calmette–Guérin (BCG) treatment as a potential modifier of phenotypic marker expression, including Nectin‑4 and TROP‑2, remains understudied in UTUC. In non‑muscle‑invasive bladder cancer (NMIBC), multiple transcriptomic analyses have shown that both Nectin‑4 and TROP‑2 expression levels generally remain stable following BCG therapy, suggesting limited direct modulation by this immunotherapy [21]. However, another molecular study reported an upregulation of these markers post‑BCG exposure in a subset of tumors [22]. These conflicting findings may be influenced by tumor subtype, treatment duration, and intratumoral heterogeneity. In our retrospective UTUC cohort, detailed histories of bladder cancer treatment—including BCG or systemic chemotherapy—were available in only a minority of cases, precluding a robust subgroup analysis. We have therefore acknowledged this as an important limitation and potential confounding factor in interpreting biomarker expression in UTUC cases with prior bladder cancer history.

Our study has several limitations. First, its retrospective design limits causal inference, making it difficult to establish definitive relationships between TROP-2 and Nectin-4 expression and clinical outcomes. Second, the small and imbalanced sample size may limit statistical power. However, this reflects the rarity of UTUC itself—which comprises only 5–10% of urothelial tumors—and the even lower incidence of coexisting UTUC and prior UBC, which constrains prospective cohort size. Third, survival data were unavailable, preventing validation of the prognostic role of TROP-2 and Nectin-4. Fourth, potential prior treatments for UB-Ca, such as intravesical BCG or chemotherapy, may have affected expression of these markers, although such treatment histories were not systematically recorded. Future research should incorporate prospective, multi-institutional cohorts with standardized treatment documentation and centralized pathology review to improve generalizability and biomarker validation.

Join us online for the 46th Virtual Global Year-End Review of Import/Export & Trade Compliance Developments Conference, a multi-panel webinar series featuring expert insights on key global trade topics.

This three-day online series features nine substantive sessions (three each day) designed to help you navigate the evolving trade landscape. Join our international trade compliance lawyers from around the world for in-depth discussions on sanctions and export controls, imports and customs compliance, enforcement, global and regional regulatory developments and more!

This virtual event will be preceded by an exclusive in-person forum in Santa Clara, California on Thursday, November 13, focused on Trade Policy in an Era of Geoeconomics, Tariff Wars & National Security Risks. To learn more and register for the in-person forum, please visit our Trade Policy in an Era of Geoeconomics, Tariff Wars & National Security Risks page.

 

Identifying critical thresholds

In this study, we performed a thorough thresholding analysis in hopes of identifying cerebrovascular reactivity thresholds that provide the most utility in deriving iICP. Beginning with PRx, a distinct threshold emerged during the Chi-square outcome analysis, + 0.05. This was the only threshold able to produce a statistically significant Chi-square for both mortality and favorable outcome prediction. Though it was unable to produce the strongest Spearman rank correlations with the cerebral physiologic insult burden measures, it was able to produce relatively strong correlations, and no other threshold was able to consistently produce the strongest correlations either. Additionally, the iICP derivation yield for + 0.05 was relatively high at 61.37%, only about 12% less than the largest yield observed. Based on the above findings, we identified + 0.05 as the likely optimal threshold for PRx-based iICP derivation. However, it is worth noting, that patients spent a relatively limited amount of time with severely elevated ICP, as shown in Table 1. This may have influenced the identification of + 0.05 as the optimal PRx threshold, as most of the data points are likely concentrated at lower ICP values and, by extension (given the relationship between ICP and PRx), at lower PRx values. Thus, the statistical strength observed at this threshold may, at least in part, reflect the underlying distribution of the dataset rather than a definitive physiologic inflection point.

For PAx-based iICP, the Chi-square analysis suggested a threshold of −0.20, as it was the threshold that produced the greatest peak when plotted. Furthermore, it was the only threshold to produce statistically significant Chi-squares when filtering for iICP with tight confidence bands, iICP.ci. However, this threshold falls within the range generally considered to represent intact reactivity (−1 to ~ 0), and is therefore, as discussed in the methods, unlikely to be a physiologically relevant threshold for iICP derivation. Although another peak was observed at + 0.55 on the mortality prediction plot, with a Chi-square only marginally smaller than that of − 0.20, this threshold was unable to produce a statistically significant Chi-square for favorable outcome prediction. It also failed to achieve statistical significance when iICP was filtered for small confidence bands, iICP.ci. Upon Spearman rank correlation testing, − 0.20 did produce the strongest correlation with any of the insult burden measures and even failed to produce a statistically significant correlation with both % time with CPP < 60mmHg and % time with CPP > 70mmHg. On the other hand, + 0.55 was able to produce the strongest correlations for these two cerebral physiologic insult burden metrics, as well as produce relatively strong associations for both % time with PRx > 0.25 and % time with RAC > 0. This seems promising for the threshold; however, + 0.55 is quite high considering that the highest identified critical PAx threshold for outcome prediction has been + 0.25 [36, 38]. Additionally, the iICP derivation yield associated with + 0.55 was abysmal at a mere 26.03%. Therefore, there appears to be no clear ideal threshold for PAx-based iICP derivation.

The Chi-square analysis for RAC-based iICP seemed to suggest − 0.45 as the ideal threshold, since it was the threshold that produced the most distinct Chi-square peak for both mortality and favorable outcome prediction. However, upon Spearman rank correlation testing, the threshold was unable to produce statistically significant correlations with % time with CPP > 70mmHg, % time with PRx > 0.25, and % time with PbtO₂ < 20mmHg. Furthermore, this threshold is well within the negative range of RAC values, and thus, unlikely represents a physiologically relevant threshold for deriving iICP. Although + 0.30 was able to produce the strongest correlations for % time with CPP < 60 mmHg, % time with CPP > 70mmHg, and % time with PRx > 0.25, it was unable to produce any statistically significant associations with outcome prediction upon Chi-square analysis. Additionally, + 0.30 was associated with an incredibly low derivation yield (16.71%), making it unviable for iICP derivation. Due to the lack of any meaningful results, no ideal threshold for RAC-based iICP could be identified. Moreover, the utility of RAC for deriving iICP remains highly unclear in general, given the complexity of interpreting this index (RAC provides insight into not only cerebrovascular reactivity but also cerebral compensatory reserve) [20]. Therefore, more work is needed to evaluate the role of RAC in deriving personalized physiologic metrics.

This study provides the first comprehensive comparison of cerebrovascular reactivity thresholds for deriving iICP. Unlike previous studies that derived iICP using an arbitrarily selected threshold, we provided an in-depth evaluation of how threshold choice influences the performance of iICP. This work will inform future works, specifically algorithm development, in threshold selection. However, it is important to clarify that the thresholds identified in this study do not represent cerebrovascular reactivity cut-off values that are themselves most predictive of outcome. Rather, they solely reflect the optimal thresholds for deriving iICP, defined as those that produced iICP values most strongly associated with clinical outcomes and multimodal cerebral physiology.

Additional findings

Through this thresholding analysis, we made multiple additional interesting observations that deserve highlighting. Firstly, the findings of this study did not completely fall in line with the critical outcome prediction thresholds identified for the three cerebrovascular reactivity indices in recent literature. This is especially true for PAx and RAC, as we were unable to identify ideal thresholds for these. The current literature suggests critical thresholds in the ranges of 0 to + 0.25 and − 0.10 to + 0.05, for PAx and RAC, respectively [36, 38]. These critical thresholds were identified through association work between the cerebrovascular reactivity indices, as stand-alone parameters (not in deriving iICP), and long-term outcome using similar chi-square analyses. It would have been reasonable to expect that these critical thresholds would have been identified as the ideal thresholds for deriving iICP as well; however, this does not seem to be the case. For PRx, the literature has generally pointed towards a critical threshold within the range of + 0.25 to + 0.35 for mortality prediction [31, 36, 38]. However, there is some literature supporting a threshold of + 0.05. In one study by Sorrentino and colleagues, + 0.05 was identified as a critical threshold for favorable outcome prediction [31]. Furthermore, there is extensive pre-clinical animal literature suggesting that a PRx around 0 detects the lower limit of autoregulation [32, 40,41,42]. These studies provide some reassurance for the PRx threshold we identified here.

Second, PAx and RAC were associated with lower iICP derivation yields when compared to PRx. This mirrors recent findings from studies comparing the three indices for CPPopt derivation [43, 44]. One possible explanation for this is that, due to the highly controlled nature of ICP in the ICU setting, there may be too little variation in AMP to produce the needed variability in these indices to generate well fitted LOESS. This may result in identification of fewer iICP values (lower yield) or identification of inaccurate iICP values, both of which can blunt the ability of iICP to predict outcome. Therefore, it is likely that PRx represents the most practical cerebrovascular reactivity index for deriving iICP, since a low yields would significantly limit any clinical utility of iICP. However, we cannot make any conclusive statements on the underlying reasoning for this difference in yields. Additionally, it is important to note that these yields were produced using the entire recording periods of patients, and that a continuous multi-window weighted approach to iICP derivation, which would be necessary for clinical application, may produce different results. We, therefore, suggest that future iICP work not exclude these indices until further work has confirmed that they are inferior to PRx for iICP derivation.

Next, during Chi-square analysis, it was observed that favorable outcome prediction tended to produce greater Chi-squares than mortality prediction for PRx-based iICP, while producing smaller values than mortality prediction for RAC-based iICP. This suggests that PRx-based iICP is better at predicting favorable outcome than predicting mortality, while RAC-based iICP is better at predicting mortality than predicting favorable outcome. Also, for mortality prediction, RAC-based iICP produced greater Chi-squares than PAx-based iICP, which produced greater Chi-squares than PRx-based iICP. On the other hand, for favorable outcome prediction, PRx- and PAx-based iICP produced greater Chi-squares than RAC-based iICP. This suggests that RAC-based iICP may be best able to predict mortality, but the worst for predicting favorable outcome.

Regarding iICP.ci, it is interesting to see that at higher thresholds, more iICP values were filtered out than at the lower thresholds (see Supplemental Appendices A-C). This suggests that at higher thresholds, confidence in the accuracy of the identified iICP diminishes. iICP.ci also generally produced greater Chi-square values for outcome prediction, but lower yields, than compared to unfiltered iICP. This suggests that iICP based on confidence band size may result in greater ability to predict outcome, but at the expense of yield. Lastly, the subgroup analysis for age and sex was unable to identify any thresholds that were able to achieve significance for the ≥ 40 age group and female group. This may potentially suggest that various patient-specific factors can affect the utility of iICP, as well as the ideal cerebrovascular threshold for its derivation. However, this finding may be a result of differences in group sizes. Further work will be needed to investigate the role that patient demographics, injury severity, and treatment regimen has on iICP derivation.

Limitations

Despite the important findings uncovered in this thresholding analysis, there are a couple noteworthy limitations that must be addressed. Firstly, the main limitation of this thresholding analysis is that we generated iICP using patients’ entire recording periods. Therefore, it remains unknown whether the ideal thresholds identified here would be applicable to a continuously derived iICP. Currently, no continuous iICP algorithm exists; however, once one is developed, a further thresholding analysis may be necessary to confirm the idealness of the identified thresholds for deriving iICP in real-time.

Another limitation of this study is that the chi-square analysis failed to produce smooth plots where the chi-square values gradually increase, peak at an “ideal” threshold, and then gradually decrease (similar to what is seen for the yield curves). Rather, the generated plots present an erratic curve with sudden spikes. This questions whether the thresholds found to produce the strongest chi-square values are physiologically significant and not just mere statistical anomalies. Future work using datasets from outside the CAHR-TBI collaborative, such as high-resolution datasets from the CENTER-TBI and TRACK-TBI studies [45, 46], will be needed to validate our findings.

Since cerebral hemodynamics exhibit significant variation throughout the different phases of post-TBI recovery, it is highly possible that the ideal thresholds for iICP derivation vary over the course of a patient’s time in the ICU [36, 47]. In this study we did not consider such variations in cerebral physiology, thus limiting our findings. Future studies should consider stratifying monitoring periods across patients’ times in the ICU to better understand how these variations in cerebral physiology may affect optimal iICP derivation.Next, though the patient cohort used in this study was quite large, only a portion of them had PbtO₂ recordings available (n = 106). This may have underpowered any tests involving this physiologic variable and may possibly explain why only one index-threshold pair was able to generate an iICP that produced a statistically significant association with % time with PbtO₂ < 20mmHg. Future work with larger PbtO₂ datasets is warranted to better shed light on the association between iICP and this important cerebral physiologic parameter.

Lastly, this study is limited by the scope of data available in the CAHR-TBI database. For instance, the database does not document whether patients underwent decompressive craniectomy, a procedure that recent literature suggests may influence cerebrovascular reactivity [48]. The absence of this information restricts our ability to account for a potentially important confounding variable. Furthermore, the lack of contemporary CT scoring systems, such as the Rotterdam or Helsinki CT scores, and the use of GOS, rather than the more detailed extended version (GOSE), limit our analyses and may affect the precision and generalizability of our findings.

Future directions

Unlike traditional static, population-based ICP thresholds, patient-specific ICP thresholds account for an individual’s dynamic cerebral autoregulatory status and may, in the future, enable treatment that is tailored to the individual’s specific physiologic needs. However, despite the promising preliminary findings regarding iICP, limited literature exists on the concept as of now. Additionally, the current state of the iICP concept is not conducive to clinical application. Firstly, the current algorithm requires a patient’s entire recording period, allowing only for the calculation of an “after the fact” threshold that is not usable to guide treatment. Moreover, it is only able to produce a singular threshold for a dataset and, therefore, does not take into account the dynamic nature of cerebral physiology over a patient’s time in the ICU. Another limitation of the current algorithm is that it fails to provide any assessment of curve fit characteristics. This prevents the clinical end-user from being able to gauge the quality of the output iICP value.

To circumvent these shortcomings, an algorithm that can continuously derive iICP in real-time is needed. Such an algorithm would require a sophisticated sliding multi-window weighted approach that, for each update interval (i.e. every minute), generates LOESS plots for various window lengths, scores plots based on a variety of factors (curve shape, confidence bands, data range, etc.), and calculates a weighted average to identify an iICP value. A similar strategy has been successfully leveraged in recent renditions of CPPopt [35, 49]. The algorithm should also present a summary of curve fit characteristics with each iICP calculation to allow for output quality assessment. Additionally, once a continuous algorithm is created, an assessment of whether the ideal CVR threshold for iICP derivation varies over different phases of the ICU stay (e.g., first 24 h vs. later periods) will be needed. This will provide valuable insight into how cerebrovascular reactivity impairment and iICP behavior evolve over time.

Following the development of such a continuous iICP derivation algorithm, thorough outcome analyses will be needed to provide preliminary insight into the prognostic utility of continuously derived iICP. Additionally, evaluation of the association between iICP and measures of cerebral physiologic insult burden will also be needed to shed light on the potential impact that iICP-directed care could have on minimizing secondary brain injury. However, to conclusively determine if iICP-directed care offers any real clinical benefit, a clinical trial would be needed.

Next, if iICP is to ever become implemented clinically, work will be needed to enable bedside implementation. This will require tailoring any continuously updating algorithm to the specific needs of the bedside environment and developing a user-interface that allows clinical end-users to efficiently use and adjust output values. Finally, while iICP represents a promising individualized approach to managing ICP, the integration of additional personalized cerebral physiologic metrics may further enhance the precision and utility of this tool. These include CPPopt, the mean arterial pressure optimum (MAPopt), and the bispectral index optimum (BISopt). In conjunction, these personalized metrics may help mitigate each other’s limitations, supporting a more comprehensive and effective strategy for bedside decision-making in neurocritical care.