Category: 8. Health

The second spot, however, is more unexpected. “The most interesting thing, I think, about the data, was that number two was the arts, entertainment and recreation… the reason that is so high is because it includes amusement parks,” says Schmid.

Agriculture, including farming, hunting, and fishing, rounds out the top three. “No surprise with hunting and farming, obviously a very hands-on industry. You’re around a lot of machinery, the same as transportation and warehousing, just a lot of heavy-duty machinery being used,” she adds.

Surprises and safety lessons

Construction, often assumed to be among the most dangerous sectors, appears lower on the list. “To see construction lower is definitely surprising. It’s got almost half the incidence rate of transportation and warehousing in number one,” Schmid observes, speculating that stricter safety regulations may be making a difference.

At the other end of the spectrum, office-based sectors like finance and insurance report the lowest rates of non-fatal injuries.

Comparing metagenomic profiles of the skin microbiome in immunosuppressed patients and those at high and low risk for cutaneous squamous cell carcinoma (SCC) revealed distinct expansions in fungal and viral taxa in the first two groups, according to the authors of a recent research letter reporting the results. Such differences one day may help clinicians identify patients at high SCC risk, they added.

Currently, physicians identify patients at high SCC risk based on whether they have had several prior SCCs. Analyzing skin microbiome changes to identify patients at high SCC risk before they develop several SCCs could revolutionize dermatologic care, senior author and investigator, Shadmehr Demehri, MD, PhD, said in an interview. He is the Arthur and Sandra Irving Endowed Chair in Cancer Immunology and director of the High Risk Skin Cancer Clinic at Massachusetts General Hospital in Boston.

Although age, ultraviolet exposure, and immunosuppression are well-known SCC risk factors, Demehri said, the skin microbiome’s contribution to SCC risk remains unclear. To explore this issue, he and his coinvestigators conducted comparative metagenomic analyses of skin microbiomes from 30 patients: six with high SCC risk (> 2 prior SCCs), nine with low SCC risk (≤ 2 prior SCCs), and 15 solid organ transplant recipients (SOTRs). The results were published online in July in the Journal of Investigative Dermatology.

Using swabs from six bilateral anatomical sites per patient, with air swabs as negative controls, the investigators performed shotgun metagenomic sequencing and differential-abundance analyses to compare relative taxa populations across patient groups. After an analysis of 249 metagenomes, high-risk individuals had the highest mean SCC count (11.8), followed by SOTRs (8.73) and low-risk individuals (0.33).

“Compared with low‐risk SCC subjects,” Demehri added, “both solid organ transplant recipients and high‐risk SCC subjects showed higher relative abundances of eukaryotes, such as Malassezia restricta and M globosa, and viruses such as Betapapillomavirus.” Moreover, these taxa showed robust differences and strong discriminatory power between patient groups, indicating their potential utility as signatures of elevated SCC risk, he told Medscape Medical News.

Which Comes First?

Currently, it remains unclear whether the relative expansion of certain skin microbiome components directly drives SCC pathogenesis or reflects underlying immune alterations that raise SCC risk indirectly. And despite the eukaryotic expansions observed in the high-risk and SOTR groups, researchers saw no between-group differences in Malassezia-related conditions such as seborrheic dermatitis. Yet subclinical expansions in Malassezia and viral taxa still may translate into increased SCC risk, Demehri said.

Accordingly, he said, dermatologists must not assume that people who develop certain viral and fungal skin diseases are at higher risk for SCC. “It’s more subtle changes, probably in association with other microbiome changes, that seem to be associated with skin cancer risk.” Nonetheless, Demehri said, watching for skin conditions that signify an expanded skin microbiome — such as the appearance of warts in fair-skinned adults — may signal the need to monitor for SCC risk.

Commensal Concern

Sancy A. Leachman, MD, PhD, professor and vice-chair of faculty development in the Department of Dermatology at the University of Utah, Salt Lake City, said that although the study requires replication in larger, prospective cohorts, its findings are provocative for several reasons. She was not involved with the study but was asked to comment.

Dermatologists are starting to embrace the skin microbiome concept but rarely beyond bacteria and yeasts, Leachman told Medscape Medical News. “I don’t believe most dermatologists are thinking about papillomaviruses being commensal: present on your skin as a normal part of your microbiome,’ she said. The observation that baseline papillomavirus populations can exist on normal-appearing skin without causing pathology is important information, she added. “What is their purpose there? What’s the evolutionary advantage?”

A study published in Cancer Cell in January 2025 showed that commensal papillomaviruses can boost immune surveillance and clearing of UV-induced p53-mutant keratinocytes. “It appears that the commensal papillomavirus population may help to stimulate the immune system in a way that helps the immune response against skin cancer,” Leachman said.

Papillomaviruses work primarily through p53. “So if you’ve developed an immune reaction against a p53 element of the papillomavirus,” Leachman said, “there’s a possibility that might cross-react with abnormalities of the p53 pathway in squamous cell carcinomas and act like a mini-vaccine against the tumor. And if that’s true, could you do that intentionally as a therapeutic or prevention strategy?”

Human Papillomavirus (HPV) vaccines have proven highly effective in preventing cervical cancer. However, she said, based on the results of commensal HPVs protecting against SCC, it is unclear whether elimination of commensals by the vaccine could render some women more or less susceptible to SCC later in life.

“If those papillomavirus vaccines cross-react, and you’re diminishing the commensal papillomaviruses that are helpful in recognizing squamous cell carcinoma,” she asked, “are you going to have people who experience an idiosyncratic increase in squamous carcinoma because the HPV vaccine prevents development of a robust commensal population of helpful papillomaviruses? I don’t believe anyone has even examined alterations of commensal HPV populations following vaccination.” 

Functional Immunosuppression

One of the study’s most tantalizing findings requiring further follow-up, Leachman said, is that the nonimmunosuppressed high-risk group (median age, 78.5 years) was much older than the low-risk group (63.0 years). “That says there’s probably some sort of functional immunosuppression occurring in those high-risk people.”

As baby boomers age, Leachman added, the population with immune-system senescence will grow. “If you can use solid organ transplant recipients as a model for the aging population that is becoming functionally immunosuppressed, that would be very beneficial, to know how to tailor treatments, detection methods, and even potential risk evaluation methods for these people.”

Anecdotally, Leachman has identified a group of patients with what she calls systemic “skin failure” — elderly patients at an extremely high SCC risk who routinely have multiple skin cancers and precancers excised. “Generally, those people are older and have an apparent functional immunosuppression; in my hands, they seem to respond better to topical imiquimod than topical 5-fluorouracil.”

Based on years of clinical observation, she prescribes topical immunotherapy rather than topical chemotherapy depending on patient age and the number and (wart-like) appearance of their SCCs.

Microbiome Manipulation?

In the interview, Demehri said that although it might be tempting to try and alter the skin microbiome through supplements or topical agents, nothing in the study suggests a cause-and-effect relationship such that modifying the microbiome would directly affect SCC risk. Therefore, he said that along with SCC detection and monitoring, immunosuppression and its role in skin cancer development should receive more emphasis.

Presently, Demehri’s lab is exploring ways to detect and monitor microbiome changes more feasibly than through costly shotgun sequencing. “It’s not something we can do for every patient all the time to monitor their skin. But there are ways we could extract that information more directly from the skin by swabs and then inform the physician that the patient might be at risk of cancer as well because their microbiome is being altered in ways that are associated with increased risk.” Broad adoption of such techniques, he estimated, is perhaps a few years away.

This study was supported partly by the Intramural Research Programs of the National Human Genome Research Institute (NHGRI) and the National Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS) at the National Institutes of Health. Study coauthors were from the Center for Cancer Immunology and the Cutaneous Biology Research Center at Massachusetts General Hospital and Harvard Medical School, Boston, and from the NHGRI and NIAMS. Demehri had filed a patent for the development of T cell-directed anticancer vaccines against commensal viruses. Other authors had no disclosures. Leachman is an investigator or advisor for several companies involved in screening and early diagnosis of melanoma but reported no relevant financial relationships.

John Jesitus is a Denver-based freelance medical writer and editor.

For more news, follow Medscape on Facebook, X (formerly known as Twitter), Instagram, and YouTube.

Data source and population

NHANES is a nationally representative survey of U.S. civilians, employing stratified multistage probability sampling. The NHANES protocols received approval from the Research Ethics Review Board of the National Center for Health Statistics (NCHS), and informed consent was obtained from all study participants. The NHANES database spans 10 survey cycles between 1999 and 2018, involving 101,316 participants who were followed up. Among these, 14,692 individuals had arthritis (https://wwwn.cdc.gov/nchs/nhanes/2009-2010/ARQ_F.htm). The inclusion criteria were (1) diagnosed with arthritis, (2) available OBS data, (3) available covariate data, and (4) available follow-up. Participants < 20 years old, pregnant women, or missing data (socioeconomic indexes, body mass index (BMI), smoking, dietary habits, or physical activity) were excluded. Particularly, in line with NHANES analytical protocols, covariates with missing values of less than 10% were directly deleted without affecting the results of the analysis, and missing variables above this threshold should be randomly interpolated and then analyzed. As shown in the flowchart, all missing values for covariates in this study were less than 10%. Therefore, individuals with missing covariates were not included in the study. Ultimately, after excluding 458 individuals with missing OBS data and 2480 with missing covariate information, this study included 11,754 patients with arthritis from the NHANES 1999–2018 (Fig. 1).

Exposure information

The data for calculating OBS scores were extracted from the NHANES database. The OBS was calculated using 16 nutrients (dietary fiber, carotene, riboflavin, niacin, vitamin B6, total folate, vitamin B12, vitamin C, vitamin E, calcium, magnesium, zinc, copper, selenium, total fat, and iron) and four lifestyle factors (physical activity, BMI, alcohol consumption, and smoking) [19]. Total fat, iron, BMI, alcohol, and smoking were considered pro-oxidant factors, while the others were considered antioxidant ones [26]. Alcohol was divided into three categories: heavy drinkers (≥15 and ≥30 g/d for women and men, respectively), drinkers (0–15 g/d and 0–30 g/d for women and men, respectively), and non-drinkers [19]. For antioxidative components, scores are assigned as 0, 1, and 2 points for the lowest to the highest tertiles, respectively. In contrast, pro-oxidative components are scored inversely, with the highest tertile receiving 0 points and the lowest receiving 2 points [19]. OBS was categorized into quartiles based on overall sample quartiles to align with prior NHANES studies and ensure sufficient sample sizes for stratified analyses [27]. The scoring scheme for OBS components is detailed in Supplementary Table S1.

Ascertainment of mortality outcome

The study outcomes were all-cause mortality and CVD mortality. All-cause mortality was defined as death from any cause. CVD mortality was defined as death from ICD-10 codes I00-I09, I11, I13, and I20-I51 [28].

Assessment of covariates

Covariates were selected a priori based on established research evidence and recommendations from clinical experts [27, 29], with the three key considerations: (1) Established confounding frameworks in arthritis mortality studies (e.g., demographics, socioeconomic status), (2) biological plausibility as mediators of oxidative stress pathways (e.g., comorbidities), and (3) empirical evidence of confounding effects. Trained staff administered structured interviews to obtain demographic data, capturing age, sex, ethnicity (non-Hispanic Black, non-Hispanic White, Mexican American, other Hispanica, and other Races), education level (less than high school, high school or equivalent, and college or above), marital status (married/cohabiting, widowed/divorced/separated, and never married), poverty income ratio (computed using HHS federal poverty standards), age at arthritis diagnosis, energy intake, disease history, alcohol use, smoking, and prescription medication use. Individuals were identified as having diabetes, hypertension, or CVD based on either having a doctor-confirmed diagnosis or being on relevant prescribed drugs. Body mass index (BMI) was measured by professionals at the Mobile Examination Centre (MEC).

Statistical analysis

Per NHANES guidelines, we applied MEC sample weights for nationally representative estimates. Normally distributed continuous variables were described using means ± standard deviations (SD) and analyzed using the analysis of variance (ANOVA) test. Non-normally distributed continuous variables were described using medians (interquartile ranges (IQRs)) and analyzed using the Wilcoxon test. The categorical variables were summarized using n (%) and analyzed using the Rao-Scott chi-square test. The Kaplan-Meier analysis was used to depict the survival rate disparities among different groups of patients, with significance determined using the log-rank test. The associations of OBS with all-cause mortality and CVD mortality were evaluated using three multivariable Cox regression models. Model 1 was the unadjusted crude model. Model 2 was adjusted for age, sex, and ethnicity. Model 3 was adjusted for age, sex, ethnicity, age at arthritis diagnosis, survey cycle, education level, marital status, PIR, energy intake, and history of diabetes or hypertension. In the multivariable Cox proportional hazard regression model, a trend test was performed across quartile groups. Restricted cubic spline (RCS) regression models, which were fitted with 3 knots at the 10th, 50th, and 90th, based on the AIC values to ensure the best fit effect, were used to explore potential non-linear associations between OBS and all-cause mortality or CVD mortality. Furthermore, sensitivity analyses were performed after excluding the patients with cancer and the patients who died within the first 2 years of follow-up to minimize reverse causality, as these groups may have pre-existing conditions influencing mortality. Stratified analyses were performed based on age (< 60 and ≥ 60 years old), sex, ethnicity, and arthritis type. All analyses were conducted using R (version 4.2.1), with statistical significance set at a P-value of < 0.05 (two-sided).

The rise of MASLD and Metabolic Syndrome poses a significant health challenge. Understanding the interplay of sociodemographic, clinical, metabolic, lipid and blood pressure factors in predicting Metabolic Syndrome among MASLD patients is crucial for effective interventions. Our study confirms previous findings and identifies new correlates, indicating the need for continued investigation.(Fig. 2)

Sociodemographic characteristics and metabolic syndrome in participants

There were no differences between gender in case of metabolic syndrome, however Patients with metabolic syndrome had a significantly higher mean age of years compared to those without metabolic syndrome in which the older age is associated with an increased risk of developing metabolic syndrome in this MASLD whereas other cross-sectional study showed no difference according to age [23]. Dyslipidemia was significantly more prevalent among individuals with metabolic syndrome, consistent with findings from a 2021 study conducted in Southwest Ethiopia [24]. There is a higher rate of diabetic retinopathy complications in metabolic syndrome patients compared to those without. This suggests that in patients with Metabolic dysfunction-associated steatotic liver disease (MASLD), the presence of diabetic retinopathy is associated with an increased risk of having metabolic syndrome, however, comparing with other studies there were no significant difference in the prevalence of metabolic syndrome between diabetics with and without diabetic retinopathy [25, 26].Research from both basic and clinical studies indicates that obesity, hypertension, hyperglycemia, hyperlipidemia, and other components of metabolic syndrome are closely interconnected and play a significant role in the onset and progression of diabetic nephropathy [27].Our study confirms this finding in which diabetic nephropathy significantly higher in the metabolic syndrome group compared to the non-metabolic syndrome patients. Also, the presence of diabetic neuropathy is associated with an increased risk of having metabolic syndrome. Insulin use is significantly higher in those compared to the non-metabolic syndrome patients. This reflects the more advanced diabetic state and insulin resistance associated with metabolic syndrome in MASLD patients.

Similar to our study findings, previous Clinical Practice Guidelines have noted that ultrasound (US) has limited sensitivity and may not accurately detect steatosis when liver fat content is below 20%, or in patients with a high body mass index (BMI) [28].

Although dapagliflozin may have a modest influence on liver enzymes [29], our study didn’t detect changes in liver enzyme levels among MS or non-MS. This suggests the therapy likely did not have any significant interference of liver enzyme markers among participants.

There is a high probability of MASLD per the HSI was seen in patients with metabolic syndrome in comparison to patients without metabolic syndrome. The HSI finding suggests that a higher degree of hepatic steatosis, is linked to an increased prevalence of metabolic syndrome in this population which also has been supported by other study [23]. On the other hand, 40.7% of non-metabolic syndrome (non-MS) patients had a high probability of MASLD based on the Hepatic Steatosis Index (HSI) due to steatosis, insulin resistance, and one or more non-MS risk factors including dyslipidemia and increased BMI. In these patients, formal diagnostic criteria for metabolic syndrome were not met, despite the presence of known components of metabolic risk. These metabolic risk factors are not unique to individuals with the diagnosis of metabolic syndrome, and contribute to the probability of having MASLD as indicated by HSI score [30].

Biomarker level and metabolic syndrome in participants

Patients with metabolic syndrome tend to have an Increased levels of systolic and diastolic pressure in comparison with others without metabolic syndrome. A study was published in 2021 Explained that MetS Patients has insulin resistance as its main component.in which insulin has an anti-natriuretic effect, and this effect can be increased n MetS Patients, which in turn can lead to hypertension within the metabolic syndrome [31]. Patients with metabolic syndrome had significantly higher mean HbA1c levels compared to those without metabolic syndrome indicating poorer glycemic control in the metabolic syndrome group. In the other hand another study revealed that higher levels of HbA1c are associated with Increased prevalence of MetS [32]. Metabolic syndrome Patients has worst lipid profile and higher levels of TG, LDL, Cholesterol and lower HDL levels, participants with MetS Patients in another study also had increased TG and decreased HDL-C which suggests that the lipid disorder had a crucial role in the development of MetS in these patients [33]. Waist circumference and BMI were significantly higher in the metabolic syndrome patients. Stolzman’s study found that adolescents with higher BMI Levels had a greater incidence of MetS than those with normal BMI [34]. In our study, we identified two novel variables, years with complication (YWC) and years since diagnosis (YSD), as significant predictors. Statistical analysis revealed that both YWC and YSD were significantly associated with biomarker levels indicative of metabolic syndrome in participants, the duration of complications and the time since diagnosis are critical factors in predicting the likelihood of metabolic syndrome in MASLD patients.

Filling a gap in the existing literature where these variables had not been previously examined.

Multivariable analysis and MASLD model

Several studies discussed the relationship between MASLD and metabolic syndrome, Yongyuan Zhang et al. confirmed the bidirectional association between MASLD and metabolic syndrome [35]. Multiple studies establish the risk of having MASLD in patient with MetS, a study published in 2022 discovered that the odds of having any level of steatosis were higher in patients with MetS [36]. Indicating that Mets increases the risk of having MASLD. Whereas a few studies focused on the Mets risk in MASLD patients, which still not fully discussed. So, we conducted a comprehensive analysis to identify significant predictors for metabolic syndrome MASLD patients. Using advanced multivariable logistic regression analysis models, the results of the analysis showed that several demographics, clinical, and metabolic factors are associated with the risk of Metabolic Syndrome in the in MASLD patients. In MASLD there is a significantly higher level of blood pressure [37], it also has been found that.

hypertension consistently exhibited the strongest link with the development of major adverse liver outcomes [38]. However, our study has found that Elevated systolic blood pressure in patients with Metabolic Dysfunction-Associated Steatotic Liver Disease (MASLD) has been linked to an increased risk of developing metabolic syndrome. Also, the increased waist circumference, will increase the risk for MetS. A study published in 2019 discovered that an increased WC is attributed to increased risk of developing DM in prediabetes with MASLD [39].

Regarding lipid profile our study pointed that higher triglycerides, and lower HDL levels were significantly linked with the metabolic syndrome outcome in MASLD patients. Anna Boulouta et al., also found that higher triglyceride and, lower HDL levels are associated with Higher risk of metabolic unhealthiness in MASLD patients [40]. Our study states result after picking up all confounding factors and found that None of the other variable -age, diabetic complications, dyslipidemia, hepatic steatosis index, HbA1c, diastolic blood pressure, BMI, LDL, total cholesterol, years since diagnosis, and years with complications showed a significant association with the presence of metabolic syndrome following rigorous adjustment for confounding factors. However other study showed that Mets risk is much less common in younger patients [40].

Lack of association between HbA1c and metabolic syndrome

Our analysis showed that HbA1c was not significantly associated with the presence of metabolic syndrome (MetS) in patients with MASLD. This finding aligns with recent evidence by Wisniewski et al. (2024) [41], who demonstrated that while HbA1c correlates with MetS components in non-diabetic individuals, this relationship disappears once type 2 diabetes mellitus (T2DM) is established. In their cross-sectional study of over 8,000 adults, they found that none of the five classical MetS criteria, including waist circumference, blood pressure, HDL-C, triglycerides, or fasting glucose, remained significantly linked to HbA1c among diabetic participants. The authors attributed this to a “glycemic ceiling effect,” whereby sustained hyperglycemia in diabetic patients narrows HbA1c variability, thereby reducing its discriminatory power for detecting metabolic clustering. In our cohort, which included only patients with established T2DM, a similar ceiling phenomenon may have occurred. This suggests that while HbA1c is essential for monitoring glycemic control, it may not serve as a reliable independent predictor of MetS once chronic dysglycemia is already present.

The use of GAM allowed us to detect potential non-linear relationships between continuous predictors and MetS. Notably, GAM revealed non-linear associations for waist circumference, HDL, systolic blood pressure, and diastolic blood pressure. These patterns were further evaluated using a multivariable logistic regression model, and the direction of associations remained consistent. This confirms that the non-linear trends captured by the GAM were not spurious and supports the robustness of the findings.

However, it is important to interpret these results with caution. Due to the cross-sectional design of this study, causal inferences cannot be made. While the identified variables show strong statistical associations with MetS, temporality and directionality cannot be determined. Thus, the findings should be viewed as correlational, highlighting variables that may warrant further investigation as potential predictors in future longitudinal or interventional studies. The model was developed in accordance with the TRIPOD (Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis) guidelines and demonstrated good discrimination and calibration (Hosmer–Lemeshow). The use of variance inflation factors (VIFs) also confirmed no significant multicollinearity between included predictors.

Our findings contribute to the growing body of literature on the metabolic burden in MASLD and offer a clinically relevant set of variables that may inform risk stratification strategies. Early identification of patients at risk of developing MetS within the MASLD population is essential, given its association with cardiovascular events, disease progression, and poor outcomes. ROC Curve for the Performance of Predictors of Metabolic Syndrome in Metabolic dysfunction-associated steatotic liver disease (MASLD) showed an Area Under the Curve (AUC) of 0.9506, which is very close to 1.0, indicating an outstanding excellent discriminative ability of systolic blood pressure, WC, TG, and HDL to predict the risk of Mets in MASLD patients, and they can very accurately distinguish between MASLD patients with and without the MetS condition.

Limitations

Several limitations merit consideration. First, the cross-sectional nature of the study limits our ability to infer temporal or causal relationships between predictors and metabolic syndrome. Second, despite incorporating a broad spectrum of clinical and biochemical variables, the possibility of residual confounding from unmeasured factors cannot be excluded. Third, as all participants were drawn from a single regional population, the generalizability of our findings to other settings or ethnic groups may be restricted. Fourth, although cardiovascular complications are of high clinical relevance in individuals with T2DM and are mechanistically intertwined with both MASLD and MetS, these could not be analyzed in our study due to non-standardized or incomplete cardiology documentation across the medical records reviewed. We therefore acknowledge this as a limitation and recommend that future prospective research include structured cardiovascular assessment to better characterize this relationship.

Conclusion and future directions

our study stated the significant predictors for metabolic syndrome using advanced statistical methods. It shows that higher systolic blood pressure, larger waist circumference, elevated triglycerides, and lower HDL cholesterol levels are independently associated with metabolic syndrome in MASLD patients. These associations were confirmed through multivariable logistic regression analysis, which accounted for potential confounding factors.

Future research should validate these findings in larger and more diverse populations and explore the underlying mechanisms of these predictors. Longitudinal studies could offer insights into causal relationships. Given the high accuracy of the GAM analysis, future studies should utilize similar advanced models to uncover non-linear relationships in clinical data, improving risk assessment tools and patient outcomes in MASLD and related conditions.

The results of the multivariate logistic regression analysis in Table 5 show that several demographics, clinical, and metabolic factors are associated with the risk of Metabolic Syndrome in the study population. The results showed that higher systolic blood pressure (adjusted OR = 1.000427, p < 0.0001) and larger waist circumference (adjusted OR = 1.001517, p < 0.0001) were both independently associated with an increased odds of having metabolic syndrome. Additionally, higher triglyceride levels (adjusted OR = 1.064834, p < 0.0001) were linked to greater odds of metabolic syndrome, while lower HDL cholesterol levels (adjusted OR = 0.998595, p = 0.003) were associated with increased odds.

The other variables, including age, diabetic complications, dyslipidemia, hepatic steatosis index, HbA1c, diastolic blood pressure, BMI, LDL, total cholesterol, years since diagnosis, and years with complications, were not significantly associated with the outcome of metabolic syndrome after adjusting for confounding factors. The Hosmer-Lemeshow test, with a chi-square statistic of 4.40 and a p-value of 0.8192, suggests that the logistic regression model fits the data well and provides an adequate representation of the observed and expected outcomes. The variance inflation factor (VIF) of 2.18 indicates that multicollinearity is not a severe issue in the regression model.

The generalized additive model analysis indicates that nonlinearity in the model is statistically significant, with a total gain (nonlinearity chi-square) of 116.313 and a p-value of 0.0000.

The generalized additive model (GAM) analysis revealed that four variables were statistically significant predictors of the binary outcome variable: waist circumference (p < 0.0001), HDL cholesterol (p < 0.0001), systolic blood pressure (p = 0.0003), and diastolic blood pressure (p < 0.0001). To further examine the potential non-linear relationships between these predictors and the outcome, we squared the values of these four variables and included them in a logistic regression model.

The results of the logistic regression confirmed that the direction of the relationships between the linear and non-linear terms for each of these four variables was consistent. This suggests that the non-linear effects of waist circumference, HDL, systolic blood pressure, and diastolic blood pressure were adequately captured in the original GAM analysis. By verifying the consistent directionality of the linear and non-linear relationships, we can have confidence that the GAM results provide an accurate representation of the underlying associations.

This approach allowed us to control for potential non-linear effects and obtain reliable estimates of the influences of these waist circumference, HDL, systolic blood pressure, and diastolic blood pressure factors on the binary outcome of interest in this population of patients with Metabolic Dysfunction-Associated Fatty Liver Disease.

Table 5 Association between predictor factors for metabolic syndrome outcome in metabolic dysfunction-associated steatotic liver disease (MASLD): A MASLD model The results of the multivariate logistic regression analysis in Table 5 show that several demographics, clinical, and metabolic factors are associated with the risk of Metabolic Syndrome in the study population. The results showed that higher systolic blood pressure (adjusted OR = 1.000427, p < 0.0001) and larger waist circumference (adjusted OR = 1.001517, p < 0.0001) were both independently associated with an increased odds of having metabolic syndrome. Additionally, higher triglyceride levels (adjusted OR = 1.064834, p < 0.0001) were linked to greater odds of metabolic syndrome, while lower HDL cholesterol levels (adjusted OR = 0.998595, p = 0.003) were associated with increased odds

The other variables, including age, diabetic complications, dyslipidemia, hepatic steatosis index, HbA1c, diastolic blood pressure, BMI, LDL, total cholesterol, years since diagnosis, and years with complications, were not significantly associated with the outcome of metabolic syndrome after adjusting for confounding factors. The Hosmer-Lemeshow test, with a chi-square statistic of 4.40 and a p-value of 0.8192, suggests that the logistic regression model fits the data well and provides an adequate representation of the observed and expected outcomes. The variance inflation factor (VIF) of 2.18 indicates that multicollinearity is not a severe issue in the regression model

The generalized additive model analysis indicates that nonlinearity in the model is statistically significant, with a total gain (nonlinearity chi-square) of 116.313 and a p-value of 0.0000

The generalized additive model (GAM) analysis revealed that four variables were statistically significant predictors of the binary outcome variable: waist circumference (p < 0.0001), HDL cholesterol (p < 0.0001), systolic blood pressure (p = 0.0003), and diastolic blood pressure (p < 0.0001). To further examine the potential non-linear relationships between these predictors and the outcome, we squared the values of these four variables and included them in a logistic regression model

The results of the logistic regression confirmed that the direction of the relationships between the linear and non-linear terms for each of these four variables was consistent. This suggests that the non-linear effects of waist circumference, HDL, systolic blood pressure, and diastolic blood pressure were adequately captured in the original GAM analysis. By verifying the consistent directionality of the linear and non-linear relationships, we can have confidence that the GAM results provide an accurate representation of the underlying associations

This approach allowed us to control for potential non-linear effects and obtain reliable estimates of the influences of these waist circumference, HDL, systolic blood pressure, and diastolic blood pressure factors on the binary outcome of interest in this population of patients with Metabolic Dysfunction-Associated Fatty Liver Disease

Table 1 presents that the study included 314 participants, with 56.4% male and 43.6% female. Most resided in cities (57.3%), followed by villages (40.1%) and camps (2.5%). MASLD was detected in 76.4% by ultrasound, with 32.8% mild, 40.1% moderate, and 3.5% severe cases. Diabetic complications included retinopathy (26.8%), nephropathy (15.0%), and neuropathy (26.1%). Dyslipidemia was present in 41.1%, and 31.2% were current smokers. Alcohol use was rare (0.3%), and no participants reported a family history of liver disease. HSI indicated a high probability of MASLD in 91.7%. Regarding treatment, 24.8% used insulin, 27.1% glimepiride, 8.9% sitagliptin, 6.4% dapagliflozin, and 82.2% were on metformin.

Table 2 show Patients with metabolic syndrome had a significantly higher mean age of 57.25 ± 10.08 years compared to those without metabolic syndrome at 53.35 ± 10.24 years (p = 0.001), suggesting that older age is associated with an increased risk of developing metabolic syndrome in this MASLD population. The prevalence of diabetic retinopathy (34.7% vs. 38.5%, p = 0.001), diabetic nephropathy (10.2% vs. 4.8%, p = 0.027), and diabetic neuropathy (17.2% vs. 8.9%, p = 0.010) was significantly higher in the metabolic syndrome group compared to the non-metabolic syndrome group, indicating that the presence of diabetic microvascular complications is linked to a higher likelihood of also having metabolic syndrome in MASLD patients. Dyslipidemia was much more common in the metabolic syndrome group, with 34.4% having dyslipidemia compared to only 6.7% in the non-metabolic syndrome group (p = 0.001), a strong association that aligns with the known components of metabolic syndrome, including atherogenic dyslipidemia. A high probability of MASLD per the HSI was seen in 50.96% of the metabolic syndrome group compared to 40.76% in the non-metabolic syndrome group (p = 0.007), suggesting that a higher degree of hepatic steatosis, as indicated by a high HSI, is linked to an increased prevalence of metabolic syndrome in this population. Insulin use was significantly higher in the metabolic syndrome group at 18.2% versus 6.7% in the non-metabolic syndrome group (p < 0.001), likely reflecting the more advanced diabetic state and insulin resistance associated with metabolic syndrome in MASLD patients.

Table 3 show Patients with metabolic syndrome had significantly higher mean HbA1c levels of 8.34% ± 1.32% compared to 7.92% ± 1.22% in those without metabolic syndrome (p = 0.004), indicating poorer glycemic control in the metabolic syndrome group. Systolic and diastolic blood pressure were also significantly elevated in the metabolic syndrome group, with median systolic BP of 136 mmHg (IQR: 130–144 mmHg) versus 125 mmHg (IQR: 118–133 mmHg) in the non-metabolic syndrome group (p < 0.001), and median diastolic BP of 86 mmHg (IQR: 82–92 mmHg) versus 82 mmHg (IQR: 76–85 mmHg) (p < 0.001). Waist circumference and BMI were significantly higher in the metabolic syndrome group, with mean values of 94.13 ± 12.10 cm and 31.17 ± 5.29, respectively, compared to 85.20 ± 6.83 cm and 27.83 ± 3.45 in the non-metabolic syndrome group (p = 0.001 for both). Lipid profiles were worse in the metabolic syndrome cohort, with higher mean LDL (129.62 ± 22.97 mg/dL vs. 105.69 ± 15.21 mg/dL, p = 0.001), lower HDL (40.69 ± 7.68 mg/dL vs. 49.76 ± 5.91 mg/dL, p < 0.001), and higher triglycerides (161.71 ± 32.83 mg/dL vs. 138.58 ± 14.33 mg/dL, p < 0.001).

Table 4 presents a comparison of clinical and biochemical characteristics between MASLD patients with and without metabolic syndrome based on the NCEP ATP III criteria. Patients with MetS were significantly older (57.6 ± 10.1 vs. 54.5 ± 9.5 years, p = 0.016). The prevalence of diabetic retinopathy and dyslipidemia was significantly higher in the MetS group (p = 0.002 and p < 0.001, respectively). A greater proportion of patients in the MetS group had a high probability of MASLD according to the Hepatic Steatosis Index (HSI) (p = 0.014). MetS patients also demonstrated significantly higher values in several cardiometabolic indicators, including systolic and diastolic blood pressure, waist circumference, BMI, LDL, triglycerides, total cholesterol, and HbA1c. In contrast, HDL levels were significantly lower among MetS patients (p < 0.001 for most comparisons). Furthermore, the MetS group had longer disease duration (YSD) and more years with complications (YWC) (p < 0.001 for both), suggesting more advanced disease and comorbidity burden.

Mutations in a single gene, HNF1A, are known to cause MODY3, a rare, early onset form of diabetes. Smaller scale mutations in the very same gene are also common and quietly nudge millions of people toward type-2 diabetes. A study published today in Cell Metabolism reveals why.

Researchers at the Centre for Genomic Regulation (CRG) in Barcelona show it’s fundamentally a problem of insulin-producing β‑cells. Using mouse models, they switched HNF1A off in different tissues and cell types including the liver, the gut and both α and β‑cells in the pancreas, one at a time. Blood glucose levels were only affected when the gene was deleted in β‑cells.

HNF1A is a known transcription factor, meaning its job is to bind to DNA and finetune the expression of other genes. The study found that deleting HNF1A in either human or mouse β‑cells affected the expression of more than one hundred genes, many of which encode for the molecular parts required to transport and release insulin.

The team also found that one of the direct targets of HNF1A happens to be A1CF, a second gene which assembles (or splices) RNA molecules before they’re turned into proteins. When HNF1A is mutated, A1CF levels collapse and the β‑cell’s RNA molecules are scrambled on a massive scale, accumulating between 1,900 and 2,300 different RNA splicing mistakes.

“When HNF1A fails, two things go wrong at once. Hundreds of genes that depend on it begin to work incorrectly. That alone is enough to weaken insulin secretion, but the loss of A1CF means that the RNAs that are still made now get spliced incorrectly. Both layers matter, but the first hit is broader and sets the stage while the second piles on extra dysfunction,” says Matías Gonzalo De Vas, co-first author of the study.

Studying human pancreatic cells painted a similar picture. In healthy donors, a robust population of β‑cells buzzed with HNF1A and A1CF activity, but in donors with type-2 diabetes, researchers observed a major increase in populations of cells with low HNF1A and A1CF activity.

“In people with type-2 diabetes, for every high-functioning β‑cell we found about eight low-functioning ones, while healthy donors had a healthier ratio of one to one. It’s a dramatic shift that shows how a single mutation can cascade into the loss of function of entire tissues and organs,” says Edgar Bernardo, co-first author of the study.

The discoveries made by the study offer a new druggable foothold for diabetes, both for MODY3, which affects around 0.03% of the general population, and type-2 diabetes, which has become so widespread that more than one in nine adults, around 600 million people worldwide, now live with the disease.

Diseases like spinal muscular dystrophy have become treatable by fixing scrambled RNA messages. Because the diabetes defect uncovered here is an RNA splicing problem, the same strategy could, in principle, be used to “re‑edit” β‑cell RNA molecules, tackling one of the root causes of the disease.

“Existing therapies for diabetes try to lower blood sugar with different strategies without correcting underlying defects. The RNA defects we found are patchable, offering a rare, clear target for an incredibly complex disease,” explains Dr. Jorge Ferrer, corresponding author of the study and researcher at the Centre for Genomic Regulation and CIBERDEM.

However, type-2 diabetes is driven by many genes and lifestyle factors. “We can now say this defective program has a causal contribution,” says Dr. Ferrer, “but there are other molecular defects that also need to be addressed. This is only one piece of a larger puzzle that we’ll also have to solve.”

His research group next plans to build what he calls a molecular parts list of the genetic chain of command, hoping to flag every possible protein and RNA molecule that could serve as a potential drug target. “The goal is to pinpoint the most practical targets for new β‑cell therapies, so we can translate these insights into effective treatments,” concludes Dr. Ferrer.

Reference: Bernardo E, De Vas MG, Balboa D, et al. HNF1A and A1CF coordinate a beta cell transcription-splicing axis that is disrupted in type 2 diabetes. Cell Metab. 2025:S1550413125003341. doi: 10.1016/j.cmet.2025.07.007

This article has been republished from the following materials. Note: material may have been edited for length and content. For further information, please contact the cited source. Our press release publishing policy can be accessed here.