The study analyzed clinical records from three dialysis centers in Chongqing, China: Daping Hospital, Kaizhou District People’s Hospital and Yongchuan Hospital of Chongqing Medical University. Eligible participants were at least 16 years old, had chronic renal failure and had received regular peritoneal dialysis for more than one month. The source cohort contained 824 records for people who began peritoneal dialysis for the first time between August 2007 and September 2025. After excluding patients with acute kidney injury, emergency dialysis, subsequent renal recovery or kidney transplantation, as well as those lost to follow-up, 737 records remained. Detection of 37 statistical outliers produced a final analytic cohort of 700 patients, who were randomly divided into a training set of 491 and a test set of 209.

During follow-up, 320 patients transitioned to hemodialysis, 178 died and 202 remained on peritoneal dialysis without either event by the end of observation. The investigators defined a hemodialysis transition as a move to hemodialysis lasting more than 90 days without a return to peritoneal dialysis, following a 2026 position statement from the International Society for Peritoneal Dialysis. Patients who died within that 90-day observation window were classified as having died rather than having made a sustained transition. Patients with unknown outcomes at censoring were treated as censored observations, not as competing deaths. Kidney transplantation and renal recovery were excluded from the primary analysis because those categories had few events, which the researchers said could make competing-risk estimates unstable.

Because the investigators expected death to compete with the planned hemodialysis transition, they compared a cause-specific proportional hazards model, two Fine–Gray model-development strategies and a conventional LASSO-Cox model that censored death. The cause-specific model used LASSO feature selection with tenfold cross-validation, while the Fine–Gray models were built using three penalized regression methods—LASSO, smoothly clipped absolute deviation and minimax concave penalty—as well as a boosting procedure. The researchers retained variables identified in univariate screening but also allowed the feature-selection procedures to consider other potentially relevant clinical characteristics. Predictive performance was assessed with time-dependent area under the curve and concordance-index curves, while calibration and discrimination were examined at three, six and nine years.

In the training set, the penalized Fine–Gray model had a mean time-dependent area under the curve of 0.716, compared with 0.696 for the cause-specific model and 0.718 for the boosted Fine–Gray model. Its training-set concordance index was 0.662. The conventional LASSO-Cox model had the highest mean training area under the curve, 0.762, but its performance was weaker in the held-out data: its test-set concordance index was 0.573, compared with 0.646 for the penalized Fine–Gray model. The penalized model also had the highest mean test-set area under the curve at 0.689. The investigators therefore selected it as the best overall model, citing its held-out performance and its smaller predictor set relative to the boosted Fine–Gray model. This was a predictive comparison within one retrospective dataset, not evidence that changing care according to the model would improve patient outcomes.

The final penalized model associated a higher body mass index with a greater subdistribution hazard of hemodialysis transition, with a subdistribution hazard ratio of 1.05 and a 95% confidence interval of 1.00 to 1.09 per BMI unit. A history of hospitalization for peritonitis was associated with more than twice the subdistribution hazard, with a ratio of 2.15 and a confidence interval of 1.64 to 2.84. Better nutrition-related parameters were associated with lower transition hazard, with a ratio of 0.27 and a confidence interval of 0.08 to 0.90. Other significant associations in the penalized model included older age, previous urological surgery and previous hemodialysis. The investigators discussed several plausible mechanisms without proving them in this study: peritonitis may make peritoneal dialysis technically unsustainable or damage the peritoneum; malnutrition may impair immune function and tissue repair; and higher BMI has previously been associated with peritonitis, loss of residual kidney function and hospitalization.

Researchers converted the penalized model into a nomogram that assigns points to clinical variables and maps a patient’s total score to estimated cumulative incidence of hemodialysis transition at three, six and nine years. In a worked example, a 60-year-old man with a BMI of 20.4 kg/m², reduced urine output and previous hospitalization for both peritonitis and cardiovascular disease received approximately 398 points. The nomogram estimated his cumulative transition risk as 39% at three years, 60% at six years and 69% at nine years. Patients in the test cohort were divided into low-, middle- and high-risk groups using score thresholds derived from the training set. Their cumulative incidence curves differed significantly for both hemodialysis transition and death, and the high-risk group had the highest incidence of transition, although a minor early crossing occurred between the low- and middle-risk groups.

Diagnostics also exposed important limitations in the model’s statistical assumptions. The proportional subdistribution hazards assumption was violated for normal 24-hour urine output, hospitalization for peritonitis and the nutrition-related parameters. Stratifying by peritonitis hospitalization did not resolve the violation, while splitting patients at a nutrition-derived threshold produced subgroups that satisfied the assumption. However, the well-nourished subgroup showed substantial differences between training and test performance: its mean area under the curve fell from 0.763 in training to 0.609 in testing, and its concordance index fell from 0.686 to 0.579. The malnutrition subgroup performed more closely between training and testing, with area-under-the-curve values of 0.714 and 0.691 and concordance indices of 0.661 and 0.640. The authors consequently treated the subgroup models as exploratory and retained nutrition as a continuous variable in the main analysis.

The retrospective design, missing laboratory data and lack of time-varying clinical measurements constrain how broadly the results can be applied. Some clinical indicators had missing rates exceeding 30%, requiring mean, nearest-neighbor or multiple-linear-regression imputation that may have introduced bias. The dataset was drawn from three centers in one Chinese city, and the paper’s limitations section describes the work as single-center despite the multicenter data sources, an internal inconsistency that further emphasizes the need for independent validation. The model also used baseline information rather than changes in nutrition, infection or laboratory status during follow-up, and it did not establish that risk prediction led to better clinical decisions or outcomes. The authors proposed prospective multicenter validation and models capable of incorporating dynamic covariates. Until those steps are completed, the penalized Fine–Gray model and nomogram are best regarded as retrospective risk-estimation tools whose discrimination is moderate, with meaningful uncertainty around their individual predictions.

Subject of Research: Competing-risk prediction of peritoneal dialysis-to-hemodialysis transition in adults

Article Title: Optimal prediction of peritoneal dialysis to hemodialysis transition: a penalized fine-gray model

Article References: Wan, J., Wei, Q., Wu, D., Hou, S., Yv, K., Tang, R., He, J., Chen, X., Li, Y., Hu, Y., Yang, J., Li, T., & Chen, K. (2026). Optimal prediction of peritoneal dialysis to hemodialysis transition: a penalized fine-gray model. Frontiers in Medicine, 13, Article 1843381. https://doi.org/10.3389/fmed.2026.1843381

Image Credits: AI Generated

DOI: 10.3389/fmed.2026.1843381

Keywords: peritoneal dialysis, hemodialysis transition, competing risks, Fine–Gray model, clinical prediction, nomogram, peritonitis, nutritional status, retrospective cohort, model validation, Optimal, prediction

The resulting rise in venous pressure can produce intracranial hypertension, cerebral edema, hemorrhage and venous infarction. Its presentation is highly variable, ranging from headache, nausea and visual disturbance to seizures, encephalopathy, focal neurological deficits and coma. Although neurovascular imaging is essential for identifying the location and extent of thrombosis, imaging alone does not describe the metabolic and physiological stress experienced by an individual patient.

The investigators analyzed data from the Henan CVST Registry at the First Affiliated Hospital of Zhengzhou University. They screened 1,108 patients treated between January 2011 and December 2025 and included 803 adults in the final analysis. Eligible patients had been admitted within 30 days of symptom onset, had available admission blood urea nitrogen and serum albumin measurements, and had follow-up data.

Patients younger than 18 years, those hospitalized more than 30 days after symptom onset and those missing laboratory or follow-up information were excluded. The study population was randomly divided into a derivation cohort of 562 patients and a validation cohort of 241. Blood samples were collected within 24 hours of hospitalization.

The blood urea nitrogen-to-serum albumin ratio, abbreviated BAR, was calculated by dividing blood urea nitrogen in milligrams per decilitre by serum albumin in grams per decilitre. Blood urea nitrogen is routinely measured and can vary with renal perfusion, intravascular volume and protein catabolism, while serum albumin is influenced by inflammation, vascular permeability, nutritional status and overall physiological reserve. Functional outcome was assessed six months after the index hospitalization by trained neurologists who were blinded to patients’ baseline BAR values and model results.

A modified Rankin Scale score of 3 to 6 was classified as a poor outcome, while scores of 0 to 2 indicated a good outcome. Only 88 of the 803 patients, or 11.0%, had a poor six-month outcome. Compared with patients who recovered with good functional status, those with poor outcomes were older, with mean ages of 45.1 and 37.4 years, respectively.

Coma at admission was reported in 68.2% of the poor-outcome group but 21.5% of the good-outcome group. Seizures and focal neurological deficits were also more frequent among patients with poor outcomes. On imaging, intracerebral hemorrhage occurred in 42.0% of the poor-outcome group compared with 15.2% of the good-outcome group, while straight sinus thrombosis was present in 42.0% and 17.5%, respectively.

These differences describe the characteristics of the analyzed groups and do not by themselves establish causes of the subsequent outcomes. The laboratory profile also separated the groups. Patients with poor outcomes had higher mean blood urea nitrogen levels, 16.3 versus 11.3 mg/dL, lower mean serum albumin concentrations, 3.7 versus 4.1 g/dL, and a higher mean BAR, 4.6 versus 2.9.

Multivariable logistic regression in the derivation cohort identified older age, coma, intracerebral hemorrhage, straight sinus thrombosis and higher BAR as independent associations with poor functional outcome. For every one-unit increase in BAR, the adjusted odds ratio for poor outcome was 1.411, with a 95% confidence interval of 1.226 to 1.625. Straight sinus thrombosis had an adjusted odds ratio of 4.525, with a 95% confidence interval of 2.267 to 9.031.

Age was also independently associated with a higher odds ratio of 1.041 per year, while coma and intracerebral hemorrhage were associated with odds ratios of 8.127 and 2.397, respectively. These estimates quantify associations in the analyzed cohort; the observational design cannot establish that BAR or any imaging feature caused the outcome. The investigators used the five independent predictors to construct a nomogram, a graphical tool that assigns points to a patient’s age, BAR, clinical status and imaging findings and converts their sum into an estimated probability of poor functional outcome.

In a nested comparison, a model containing the clinical and radiological predictors without BAR had an AUC of 0.680 in the derivation cohort and 0.764 in the validation cohort. Adding BAR raised those values to 0.900 and 0.893, respectively, indicating better discrimination within this dataset. The BAR-containing nomogram also had Brier scores of 0.068 in the derivation cohort and 0.076 in the validation cohort, with calibration curves showing close agreement between predicted and observed risks across probability ranges.

The comparison indicates that BAR added predictive information to the measured clinical and radiological variables, but it does not establish that the nomogram improves treatment decisions or patient outcomes. Decision-curve analysis suggested that the BAR-containing model provided greater net benefit than the model without BAR across a range of clinically meaningful threshold probabilities in both cohorts. Additional analyses of the full sample recovered the same five independent predictors.

Ten-fold cross-validation and 1,000 bootstrap resampling iterations also produced stable optimism-corrected estimates for discrimination, Brier score and calibration slope. The analysis therefore supports the internal consistency of the model, but it does not show that using the nomogram would improve treatment decisions or patient outcomes. The study evaluated prediction, not the effect of a clinical intervention, and the favorable AUC values should be interpreted as performance within the study’s data rather than as a guarantee of accuracy in other hospitals or populations.

Several limitations constrain how far the findings can be generalized. The model has not yet been tested in independent cohorts from other institutions or healthcare systems, so its external calibration and clinical usefulness remain uncertain. Sampling was retrospective, and blood collection within 24 hours of admission did not standardize the interval between symptom onset, treatment and blood drawing.

Hydration changes and early therapeutic interventions may therefore have influenced blood urea nitrogen or albumin measurements. The retrospective design is also vulnerable to selection bias, despite prespecified exclusions, and the relatively small number of poor outcomes, 88 in the full sample, limits statistical precision. The findings should accordingly be regarded as an internally validated association and prediction model awaiting multicenter external testing.

Even with those constraints, the study points to BAR as a potentially useful addition to imaging-based assessment in CVST. Because both laboratory measurements are already obtained in routine hospital care, BAR may help translate existing information into a more individualized risk estimate.

Subject of Research: Prognostic value of the blood urea nitrogen-to-serum albumin ratio with clinical and radiological features in cerebral venous sinus thrombosis

Article Title: Prognostic model integrating the blood urea nitrogen-to-serum albumin ratio and radiological features for poor functional outcome in cerebral venous sinus thrombosis

Article References: Zhao, J., Liu, K., Li, S., Liu, H., Xu, Y., Xia, Z., & Song, B. (2026). Prognostic model integrating the blood urea nitrogen-to-serum albumin ratio and radiological features for poor functional outcome in cerebral venous sinus thrombosis. Frontiers in Neurology, 17, Article 1953972. https://doi.org/10.3389/fneur.2026.1953972

Image Credits: AI Generated

DOI: 10.3389/fneur.2026.1953972

Keywords: cerebral venous sinus thrombosis, blood urea nitrogen, serum albumin, prognostic model, nomogram, straight sinus thrombosis, intracerebral hemorrhage, functional outcome, clinical biomarkers, internal validation, Prognostic, model

The analysis used data collected in 2026 at the Huayuan Road Community Health Service Center in Haidian District, Beijing. Eligible participants were at least 45 years old, met the study’s abdominal-obesity criterion of a waist circumference of at least 90 cm for men and 85 cm for women, and could complete the required questionnaires and physical-fitness tests. From an original dataset of 308 community residents, 212 were included: nine were excluded because they were younger than 45, 86 because they did not meet or could not be classified by the abdominal-obesity criterion, and one because Pittsburgh Sleep Quality Index data were missing.

The final group had a mean age of 65.81 years, and 152 participants, or 71.7%, were women. Current poor sleep quality was defined as a score above 5 on the validated Chinese version of the Pittsburgh Sleep Quality Index, which assesses sleep during the preceding month through seven components: subjective sleep quality, sleep latency, sleep duration, habitual sleep efficiency, sleep disturbances, use of sleeping medication and daytime dysfunction. Scores range from 0 to 21, with higher values indicating poorer sleep.

Of the 212 participants, 117, or 55.2%, met the poor-sleep threshold, while 95 had scores of 5 or lower. The 55.2% proportion describes this selected sample and should not be treated as a prevalence estimate for all community-dwelling adults with abdominal obesity. Ten candidate predictors were specified before model development: age, sex, depressive symptoms measured with the 9-item Patient Health Questionnaire, or PHQ-9; anxiety symptoms measured with the 7-item Generalized Anxiety Disorder scale, or GAD-7; global cognition measured with the Mini-Mental State Examination; and five tests representing upper- and lower-limb strength, flexibility, cardiorespiratory endurance and functional mobility.

The physical measures were grip strength, repetitions completed in a 30-second chair-stand test, chair sit-and-reach distance, steps completed in a two-minute step test and time in a modified 8-foot up-and-go test. Least absolute shrinkage and selection operator, or LASSO, logistic regression with repeated stratified 10-fold cross-validation retained age, PHQ-9, GAD-7, grip strength, chair-stand repetitions, sit-and-reach distance and two-minute step count. The seven selected variables were entered simultaneously into an unpenalized multivariable logistic regression, from which the researchers constructed the nomogram.

Predictor-specific points are added to produce an estimated probability of current poor sleep quality. In the study’s illustration, a 65-year-old participant with a PHQ-9 score of 8, a GAD-7 score of 8, grip strength of 30 kg, 16 chair stands, a sit-and-reach distance of 0 cm and 100 steps in the two-minute test received approximately 103.2 points, corresponding to an estimated 91% probability of poor sleep quality. This is an example calculated from the development model, not a diagnosis and not an individually validated prediction.

Psychological symptoms produced the clearest individual statistical association in the multivariable analysis. Each one-point increase in PHQ-9 score was associated with 37.4% higher adjusted odds of current poor sleep quality, with an odds ratio of 1.374 and a 95% confidence interval of 1.164 to 1.621. The adjusted associations for age, GAD-7 score, grip strength, chair-stand repetitions, sit-and-reach distance and two-minute step count did not reach conventional statistical significance.

That does not mean the complete model contains no information from those variables: LASSO selects predictors according to their contribution to penalized model fit and prediction error rather than an individual p-value threshold, and prediction models are evaluated primarily through discrimination, calibration and overall prediction error. Participants classified as having poor sleep quality were older and had higher PHQ-9 and GAD-7 scores than those with better sleep. They also had lower grip strength, fewer chair-stand and arm-curl repetitions, shorter sit-and-reach distances, shorter eyes-closed single-leg stance times and fewer repetitions in the two-minute step test.

These are cross-sectional group differences, not evidence that poor sleep caused reduced physical capacity or psychological symptoms. Reciprocal links involving fatigue, physical activity, mood, inflammation and metabolic regulation are possible biological contexts discussed in the research, but the study did not directly test those mechanisms. Its data therefore cannot establish temporality, causality or the direction of any relationship.

After the seven-predictor structure was fixed, 1,000 bootstrap resamples were used to estimate and correct performance optimism. The corrected AUC was 0.751, compared with an apparent value of 0.779. The corrected calibration slope was 0.858, the calibration intercept was 0.002 and the Brier score was 0.206.

The routine-information comparator contained age, sex, number of disease categories, current alcohol consumption and exercise level. Its corrected AUC was 0.608 and its corrected Brier score was 0.245. The apparent AUC difference was 0.131, with a 95% confidence interval of 0.050 to 0.213 and a paired DeLong test p-value of 0.002.

Decision-curve analysis indicated greater theoretical net benefit for the full model across threshold probabilities of 13% to 80% in the development sample, but those thresholds are sample-dependent and do not establish a clinical or screening cutoff. Several limitations sharply constrain interpretation. A post hoc assessment using the Riley framework estimated a minimum of 381 participants, including approximately 211 outcome events, for conservative prediction-model development, so the available sample did not meet all adequacy criteria.

Bootstrap correction did not repeat the entire variable-selection process, meaning some optimism cannot be excluded. Sleep was assessed only by self-report, without actigraphy or polysomnography, and PHQ-9 item 3 asks about sleep disturbance. Because item-level PHQ-9 data were unavailable, the researchers could not remove that overlapping item to test whether it inflated the PHQ-9 association, model performance or the apparent advantage over routine information.

The findings support a possible role for opportunistic triage in community healthy-aging, obesity-management or functional-assessment programs where psychological questionnaires and field-based fitness tests are already collected. In such settings, the model could identify people for whom direct administration of the PSQI or a clinical sleep assessment may be useful, rather than replace direct sleep screening. It would have no clear advantage when the PSQI can already be obtained easily.

Independent external validation in geographically diverse populations, evaluation with objective sleep measures and assessment of staff training, referral pathways and access to sleep services remain necessary. Until those steps are completed, the nomogram is best regarded as an exploratory research tool rather than a basis for individual treatment decisions or public-health policy.

Subject of Research: Internal validation of a nomogram using psychological and physical-fitness indicators to identify current poor sleep quality in adults with abdominal obesity

Article Title: Community-based identification of poor sleep quality in middle-aged and older adults with abdominal obesity using psychological and physical fitness indicators

Article References: Qiao, C., Yang, S., Wang, Y., Yang, C., He, Y., & Zeng, S. (2026). Community-based identification of poor sleep quality in middle-aged and older adults with abdominal obesity using psychological and physical fitness indicators. Frontiers in Public Health, 14, 1964356. https://doi.org/10.3389/fpubh.2026.1964356

Image Credits: AI Generated

DOI: 10.3389/fpubh.2026.1964356

Keywords: poor sleep quality, abdominal obesity, prediction model, physical fitness, depressive symptoms, anxiety symptoms, healthy aging, community health, internal validation, nomogram, Community-based, identification

The researchers analyzed 248 neonates admitted to the neonatal intensive care unit of Southwest Hospital between January 2018 and January 2024. Seventy-seven children had NEC, defined as Modified Bell stage II or III, and 171 controls had stage I disease or other gastrointestinal manifestations, including vomiting, regurgitation or abdominal distension, without progression to NEC. The sample included 14.1% born before 32 weeks of gestation and 13.3% with a birth weight below 1,500 g.

Candidate predictors comprised perinatal characteristics, feeding-related variables, pre-symptom clinical-course information and routine laboratory indicators measured within 48 hours after birth. Multivariable logistic regression was used to construct the nomogram, which ultimately included the 5-minute Apgar score, parenteral nutrition, feeding-advancement rate, neutrophil percentage and procalcitonin. At an optimal cutoff of 0.353, the model had sensitivity of 71.4% and specificity of 86.0%.

Model assessment included discrimination, calibration, decision-curve analysis and bootstrap internal validation. Calibration analysis indicated acceptable agreement between predicted and observed risks, while decision-curve analysis suggested potential clinical usefulness. These findings support the model’s internal performance rather than establishing clinical benefit. The authors noted that external validation, particularly in very-preterm and very-low-birth-weight populations, is required before clinical implementation. The retrospective, single-center design and the selected population of neonates with suspected NEC or related gastrointestinal manifestations limit generalizability and mean that the model was not evaluated as a universal prediction tool for all newborns.

Subject of Research: Internal validation of an early necrotizing enterocolitis risk-stratification nomogram in neonates with gastrointestinal manifestations

Article Title: Early risk stratification of necrotizing enterocolitis using 48-h laboratory indicators and pre-symptom clinical-course variables: a single-center retrospective study

Article References: Lang, Y., Hu, Y., Yan, L., & Shen, L. (2026). Early risk stratification of necrotizing enterocolitis using 48-h laboratory indicators and pre-symptom clinical-course variables: a single-center retrospective study. BMC Pediatrics. https://doi.org/10.1186/s12887-026-07665-x

Image Credits: AI Generated

DOI: 10.1186/s12887-026-07665-x

Keywords: necrotizing enterocolitis, neonatal intensive care, risk stratification, nomogram, laboratory indicators, feeding intolerance, premature infants, internal validation, Early, risk, stratification, necrotizing

Researchers reviewed patients treated at four Chinese centers from October 2019 through December 2025. One center supplied a training cohort of 546 patients, while three independent centers supplied an external validation cohort of 344. Disease-free survival events occurred in 172 training participants and 93 external-validation participants. Predictors were selected using least absolute shrinkage and selection operator-penalized Cox regression and bootstrap stability selection. The resulting nomogram incorporated ypN stage, perineural invasion, major pathological response, ypT stage, sex, body mass index category and lymphovascular invasion.

The model’s discrimination was 0.741 (95% confidence interval, 0.702–0.780) in the training cohort and 0.701 (0.644–0.757) in external validation. Optimism-corrected C-indices were 0.726 (0.688–0.762) for fixed-model validation and 0.706 (0.668–0.747) when predictor selection was repeated. In the external cohort, the C-index difference between the nomogram and categorical ypTNM staging was 0.048 (0.014–0.082). External calibration showed reasonable agreement between predicted and observed disease-free survival at one and two years, but the model overpredicted disease-free survival at three years. Decision-curve analysis indicated greater net benefit than ypTNM staging within specified threshold ranges.

Using a fixed cutoff derived from the training cohort, patients classified as high risk had poorer disease-free survival in both the training cohort (hazard ratio, 3.86; 95% confidence interval, 2.75–5.44) and the external validation cohort (hazard ratio, 3.40; 95% confidence interval, 2.16–5.35; both P<0.001). The results support further evaluation of the nomogram for individualized postoperative risk estimates among patients with ESCC treated with neoadjuvant chemoimmunotherapy followed by surgery. Because the evidence presented here comes from a retrospective study and an early-access abstract, the reported performance does not establish clinical benefit, and prospective validation remains necessary.

Subject of Research: External validation of a postoperative disease-free survival prediction model in esophageal squamous cell carcinoma

Article Title: Development and external validation of a nomogram to predict disease-free survival after neoadjuvant chemoimmunotherapy and surgery in esophageal squamous cell carcinoma: a multicenter retrospective study

Article References: Ao, Y., Leng, C., Shao, L., Liu, G., Li, K., Chen, J., Huang, Y., Luo, K., Liu, Q., Zhao, L., Peng, S., & Fu, J. (2026). Development and external validation of a nomogram to predict disease-free survival after neoadjuvant chemoimmunotherapy and surgery in esophageal squamous cell carcinoma: a multicenter retrospective study. BMC Cancer. https://doi.org/10.1186/s12885-026-17164-y

Image Credits: AI Generated

DOI: 10.1186/s12885-026-17164-y

Keywords: esophageal squamous cell carcinoma, neoadjuvant chemoimmunotherapy, esophagectomy, disease-free survival, prognostic nomogram, external validation, retrospective cohort, ypTNM staging, Development, external, validation, nomogram

Neonatal necrotizing enterocolitis, commonly called NEC, is the most common severe gastrointestinal emergency among preterm and very-low-birth-weight infants, according to the study’s background. It involves bowel-wall ischemia, inflammation and necrosis, often has nonspecific early symptoms and can progress rapidly. Conventional abdominal radiography may not reveal subtle bowel-wall changes during early disease. Abdominal ultrasound is noninvasive, repeatable and free of radiation, and it can display bowel-wall thickness, intestinal movement, blood perfusion and subtle intra-abdominal changes. The researchers therefore tested whether a visual prediction model combining laboratory inflammation markers with multiparametric ultrasound could quantify an infant’s risk of early NEC.

The participants were 500 infants born before 37 weeks’ gestation who had been admitted to the neonatal intensive care unit of Quanzhou Maternal and Child Health Hospital in China within 24 hours of birth. Eligibility also required complete clinical, laboratory and ultrasound data. The investigators excluded infants with congenital gastrointestinal malformations, congenital heart disease or inherited metabolic diseases, as well as those with severe birth asphyxia, severe intracranial hemorrhage or surgery after admission. Stays shorter than 72 hours and loss to follow-up were also excluded. NEC was diagnosed using modified Bell staging criteria, with clinical findings, abdominal ultrasound and radiographic examinations contributing to the determination. All cases were stage I or II; severe stage III disease was excluded, making the model specific to early disease.

Within the retrospective cohort, infants were stratified by whether they developed NEC and then randomly allocated in a 7:3 ratio to development and validation groups. The development cohort contained 350 infants, including 68 who developed NEC and 282 who did not. The validation cohort contained 150 infants, of whom 29 developed NEC and 121 did not. Altogether, 97 of the 500 infants developed early-stage NEC. The two cohorts were internal subsets of the same single-center sample rather than independent populations recruited from different hospitals. Clinical variables were taken from electronic medical records, and C-reactive protein, procalcitonin and interleukin-6 were measured in peripheral venous blood collected within 72 hours of admission.

Two senior ultrasonographers performed color Doppler examinations without access to clinical manifestations or laboratory results. They used a probe frequency of 7.5 to 10.0 MHz and recorded bowel-wall thickness, intestinal peristalsis, gastric antral area, gastric antral peristalsis and bowel-wall blood perfusion. Bowel-wall thickening was defined as a thickness of at least 2 mm, reduced intestinal peristalsis as fewer than three movements per minute, enlarged gastric antral area as more than 1.5 cm², and reduced gastric antral peristalsis as fewer than two movements per minute. Intramural blood flow was categorized as normal, sparse, absent or overperfused, with the latter three grouped as abnormal for modeling. Inter-rater reliability was assessed in 80 randomly selected cases. Intraclass correlation coefficients were 0.84 for bowel-wall thickness and 0.81 for gastric antral area, while Cohen’s kappa values ranged from 0.76 to 0.80 for the categorical ultrasound measures.

In the development cohort, infants who developed NEC had a mean gestational age of 31.17 ± 2.21 weeks, compared with 34.94 ± 2.03 weeks among those who did not. Mean birth weight was 1,442.74 ± 364.21 grams in the NEC group and 2,197.43 ± 428.17 grams in the non-NEC group. Mean C-reactive protein was 18.81 ± 5.47 mg/L versus 6.14 ± 2.20 mg/L; mean procalcitonin was 3.94 ± 1.31 ng/mL versus 0.82 ± 0.33 ng/mL; and mean interleukin-6 was 158.74 ± 47.13 pg/mL versus 57.41 ± 18.62 pg/mL. Abnormal bowel-wall perfusion occurred in 94.12% of infants with NEC and 19.15% of those without it. Sex, delivery mode and Apgar scores did not differ significantly between the groups in the univariate analysis.

Multivariable logistic regression identified nine independent associations with early NEC: gestational age, C-reactive protein, procalcitonin, interleukin-6, bowel-wall thickening, reduced intestinal peristalsis, enlarged gastric antral area, reduced gastric antral peristalsis and abnormal bowel-wall perfusion. Birth weight was associated with NEC in the univariate analysis but was excluded from the final model because of strong collinearity with gestational age; their variance inflation factors were 7.82 and 8.14. The nine retained predictors had variance inflation factors between 1.21 and 1.48. Each additional week of gestational age was associated with lower odds of early NEC, with an odds ratio of 0.831 and 95% confidence interval of 0.765–0.902. The odds ratios for the three laboratory markers were 1.242 for C-reactive protein, 1.385 for procalcitonin and 1.219 for interleukin-6.

The five ultrasound variables also remained associated with early NEC after adjustment. The model’s odds ratio was 1.572 for bowel-wall thickening, 8.215 for reduced intestinal peristalsis, 2.452 for enlarged gastric antral area, 7.265 for reduced gastric antral peristalsis and 14.456 for abnormal bowel-wall perfusion. The researchers converted the nine regression coefficients into a nomogram with separate scoring axes, a total-score axis and a risk axis for early NEC. Although the tool is described as ultrasound-based, it also requires gestational age and three laboratory measurements, making it a multimodal model rather than a stand-alone imaging test. Sensitivity analyses that removed one inflammatory marker at a time left the direction and statistical significance of the remaining associations unchanged, while the development-cohort AUC varied only between 0.861 and 0.883.

When the Youden index was used in the development cohort, 0.18 was selected as the optimal predicted-risk threshold and then applied without adjustment to the validation cohort. At this cutoff, the development cohort produced 82.4% sensitivity, 81.2% specificity, 51.4% positive predictive value and 95.0% negative predictive value. In validation, the corresponding measures were 77.2%, 76.0%, 43.8% and 92.4%. Thus, the positive predictive value was lower in validation: among infants flagged as positive, fewer than half developed early NEC in that cohort. By contrast, a negative result was associated with a high probability that early NEC did not occur during follow-up. Calibration was favorable, with goodness-of-fit chi-square values of 1.235 and P = 0.998 in development and 1.642 and P = 0.995 in validation. Decision-curve analysis suggested net benefit over treating all or none when the probability threshold exceeded 0.05.

Several limitations constrain the findings. As a retrospective, single-center study, the model is susceptible to selection bias and may not generalize to hospitals with different patient populations, resources or ultrasound practices. Although the investigators planned for at least 10 NEC events per predictor, the development cohort contained approximately 7.5 events per predictor, below the conventional logistic-regression benchmark. Feeding patterns, intestinal microbiota and umbilical arterial catheterization were not included, and no formal correction for multiple comparisons was applied during univariate screening. Stage I cases represent suspected NEC with limited diagnostic specificity, which may have affected the estimates. Follow-up lasted 28 days and could therefore miss later events, while ultrasound measurements remain partly operator-dependent and were assessed for inter-rater reliability in only 80 infants. Sparse, absent and hyperperfused perfusion were also collapsed into one binary category.

Overall, the study shows that gestational age, inflammatory markers and multiparametric ultrasound findings can be combined into a quantitative model with favorable internal predictive performance for early-stage NEC in preterm infants. The lower validation positive predictive value indicates that the tool should not be interpreted as a definitive diagnosis, and the research did not test whether nomogram-guided care reduces missed diagnoses, disease progression or mortality. Prospective, multicenter external validation is still needed to establish performance in other neonatal intensive care units and to determine whether the model can support decisions in practice. Until then, the results are best viewed as evidence that bedside ultrasound and conventional laboratory information can complement one another in neonatal risk assessment, not as proof of a new diagnostic standard.

Subject of Research: Internal development and validation of a nomogram predicting early-stage neonatal necrotizing enterocolitis in preterm infants using inflammatory markers and abdominal ultrasound.

Article Title: An ultrasound-based nomogram for predicting early neonatal necrotizing enterocolitis

Article References: RUAN, J., SU, H., HE, Y., & LIN, W. (2026). An ultrasound-based nomogram for predicting early neonatal necrotizing enterocolitis. Frontiers in Pediatrics, 14, Article 1907406. https://doi.org/10.3389/fped.2026.1907406

Image Credits: AI Generated

DOI: 10.3389/fped.2026.1907406

Keywords: neonatal necrotizing enterocolitis, preterm infants, abdominal ultrasound, clinical nomogram, risk prediction, inflammation markers, bowel perfusion, internal validation, neonatal intensive care, ultrasound-based, nomogram, predicting

The cohort comprised patients admitted to the Pain Department at Hengshui People’s Hospital in China between December 2020 and December 2023. Acute herpetic neuralgia was defined as pain occurring within 30 days after the appearance of a herpes zoster rash, with a Numerical Rating Scale score above 4. Eligible patients were at least 20 years old and had complete clinical and laboratory data from the index admission. The researchers excluded people with previous VTE, current anticoagulant or antiplatelet treatment, severe hepatic or renal dysfunction, known active malignancy, recent chemotherapy or radiotherapy, pregnancy or breastfeeding, and several unstable systemic conditions. Consequently, the findings describe a selected clinical population rather than all people with herpes zoster or neuralgia.

VTE outcomes were restricted to events occurring during the index hospitalization and confirmed by objective imaging. Deep vein thrombosis was diagnosed with compression or duplex ultrasonography, while pulmonary embolism was identified through computed tomography pulmonary angiography or, when that procedure was contraindicated, ventilation-perfusion scanning. Elevated D-dimer and clinical suspicion alone were not sufficient for diagnosis. The investigators collected demographic characteristics, comorbidities, laboratory measurements, pain scores and an admission thrombosis risk score from the first blood samples drawn after admission, before VTE diagnosis and before anticoagulant therapy. The thrombosis score was an institutional adaptation of the Caprini model, recorded as an aggregate from 0 to 8; individual Caprini items were not systematically documented.

Patients who developed VTE differed from those who did not in several baseline measures. Their median age was 75 years, compared with 65 years in the non-VTE group. Median fibrin degradation product concentration was 4.3 mg/L in the VTE group and 1.9 mg/L among those without VTE. Median Numerical Rating Scale scores were 8 and 7, respectively, and the median institutional thrombosis risk score was three in the VTE group and two in the non-VTE group. Elevated D-dimer, defined as a value above 0.5 mg/L fibrinogen-equivalent units, was present in 57.7% of patients with VTE and 13.3% of those without an event. These were observed group differences in a retrospective cohort, not evidence that any individual factor caused thrombosis.

To select candidate predictors, investigators used least absolute shrinkage and selection operator regression, commonly called LASSO, with 10-fold cross-validation to guide the regularization process. Ten variables were retained at the selected stage: sex, age, rheumatism and immunity, D-dimer, activated partial thromboplastin time, fibrinogen, fibrin degradation products, antithrombin III, Numerical Rating Scale score and thrombosis risk score. These variables were entered into multivariable logistic regression and refined through forward and backward stepwise selection based on the Akaike information criterion. Age correlated strongly with the institutional thrombosis risk score, at r = 0.627, because that assessment incorporated age; the age-related information was therefore largely captured by the score during variable selection. Fibrin degradation products had the strongest correlation with elevated D-dimer, at r = 0.582, and a weaker correlation with fibrinogen, at r = 0.357. Variance inflation factors for the four final predictors ranged from 1.018 to 1.534, indicating no meaningful multicollinearity within the model.

The final logistic regression associated three variables with higher adjusted odds of VTE and one with lower adjusted odds. Each 1 mg/L increase in fibrin degradation products was associated with an odds ratio of 1.48, with a 95% confidence interval of 1.282–1.737 and P less than 0.001. A one-point increase in the Numerical Rating Scale corresponded to an odds ratio of 1.584, with a 95% confidence interval of 1.05–2.538 and P = 0.039. Each additional point in the institutional thrombosis risk score was associated with an odds ratio of 2.05, with a 95% confidence interval of 1.455–2.947 and P less than 0.001. Fibrinogen was associated with lower odds, with an odds ratio of 0.477 per g/L, a 95% confidence interval of 0.275–0.798 and P = 0.006. These estimates describe conditional associations in the fitted model rather than interventions that would change a patient’s risk.

The inverse fibrinogen association requires particular caution. Fibrinogen was not significantly associated with VTE in the unadjusted analysis, with an odds ratio of 1.025 per g/L, a 95% confidence interval of 0.631–1.516 and P = 0.912. The inverse pattern emerged only after adjustment for fibrin degradation products and other correlated coagulation markers, which the authors considered more consistent with statistical suppression or negative confounding than with a directly protective biological effect. They also noted that fibrinogen is an acute-phase reactant and a substrate for fibrin formation, while citing inconsistent findings from previous studies about its relationship with VTE. Admission samples were obtained before VTE diagnosis and anticoagulant use, making treatment-related distortion unlikely. Likewise, severe pain might accompany reduced mobility and impaired venous flow, but bed-rest time and mobility status were not systematically recorded, so that proposed pathway could not be tested directly.

The four retained predictors were translated into a graphical nomogram that assigns points for fibrinogen, fibrin degradation products, Numerical Rating Scale score and the admission thrombosis risk score. Adding the assigned points yields an estimated probability of VTE during the index hospitalization. In the full cohort, the model’s apparent AUC was 0.853. Ten-fold cross-validation produced a C-index of 0.849, although its 95% confidence interval was wide, from 0.591 to 1.000, reflecting the limited number of events. Bootstrap validation with 1,000 repetitions yielded an optimism-corrected C-index of 0.834; the bootstrap distribution had a mean of 0.831 and a 95% confidence interval of 0.680–0.950. Because the model was developed on the full cohort of 327 patients rather than in a separate training sample, these results represent internal validation rather than evidence of performance in an independent population.

At the Youden-index cutoff of 0.094, the model identified 19 true positives and seven false negatives among the 26 patients with VTE, alongside 264 true negatives and 37 false positives among the 301 without VTE. Sensitivity was 0.731, with a 95% confidence interval of 0.522–0.884, and specificity was 0.877, with a 95% confidence interval of 0.835–0.912. Positive predictive value was 0.339, whereas negative predictive value was 0.974; overall accuracy was 0.865. The modest positive predictive value is consistent with the low base rate: most patients flagged as positive would not develop VTE. Calibration appeared close to the observed frequencies, with a Hosmer–Lemeshow chi-square value of 5.848 and P = 0.664, while decision-curve analysis indicated net benefit across most threshold probabilities.

Important limitations prevent the nomogram from being treated as a stand-alone diagnostic test. The retrospective, single-center design limits generalizability, and only 26 events occurred, creating substantial overfitting risk despite LASSO regularization and internal validation. Imaging was performed when clinically indicated rather than through universal screening, so asymptomatic events may have been missed and the reported incidence may be underestimated. Pharmacologic and mechanical prophylaxis was not systematically recorded, and residual confounding by prophylaxis therefore could not be excluded. Mobility data were unavailable, and the model could not be benchmarked against formally reconstructed Caprini, Padua or IMPROVE/IMPROVE-DD scores. The findings are therefore hypothesis-generating: larger multicenter prospective studies with standardized risk assessments, prophylaxis records and mobility data are needed before the model can be evaluated for routine clinical use.

Subject of Research: Development and internal validation of an in-hospital venous thromboembolism prediction model for adults with acute herpetic neuralgia

Article Title: Analysis of risk factors for venous thromboembolism in hospitalized patients with acute herpetic neuralgia and construction of a predictive nomogram model

Article References: Wang, C., Wang, G., Liu, B., Gao, Q., Du, C., & Li, L. (2026). Analysis of risk factors for venous thromboembolism in hospitalized patients with acute herpetic neuralgia and construction of a predictive nomogram model. Frontiers in Neurology, 17, Article 1877838. https://doi.org/10.3389/fneur.2026.1877838

Image Credits: AI Generated

DOI: 10.3389/fneur.2026.1877838

Keywords: acute herpetic neuralgia, venous thromboembolism, deep vein thrombosis, pulmonary embolism, prediction model, nomogram, fibrin degradation products, pain intensity, thrombosis risk score, clinical validation, Analysis, risk

The research, led by Tao Song, Yiwen Wang, Jieni Ding, Huafeng Shou, Yunwang Chen, Hongzhi Zhang, and senior author Hong’en Xu across hospitals affiliated with Hangzhou Medical College, Zhejiang University School of Medicine, and other institutions in eastern China, set out to build a predictive tool that combines clinical information with quantitative image features extracted from computed tomography. The team focused on a specific anatomic landmark: the psoas muscle at the level of the third lumbar vertebra. This large muscle flanking the spine is a common proxy for whole-body muscle status in imaging research, and the investigators reasoned that its texture, shape, and density on CT might encode information about a patient’s physiological reserve, including how well her bone marrow and immune system would tolerate the assault of radiation and chemotherapy.

The technical foundation of the study is radiomics, a field that converts medical images into high-dimensional data. Rather than asking a radiologist to eyeball the scan, radiomics pipelines extract thousands of numerical features automatically: first-order statistics describing the distribution of pixel intensities, texture metrics capturing patterns of spatial heterogeneity, and shape descriptors quantifying geometry. In this study, the team analyzed 2,153 such features from the planning CT scans of 150 patients with surgically treated stage IB-IIA cervical cancer. All patients subsequently received postoperative radiotherapy or chemoradiotherapy, and the researchers tracked which of them developed moderate-to-severe leukopenia as an adverse event.

Extracting 2,153 features from 150 patients creates a classic statistical dilemma: far more variables than outcomes, a situation that creates a substantial risk of overfitting. The Chinese team addressed this with least absolute shrinkage and selection operator regression, universally abbreviated as LASSO, a machine learning technique that performs feature selection and regularization simultaneously. LASSO works by penalizing the absolute size of regression coefficients, effectively shrinking most of them to zero and retaining a smaller set of features under the chosen penalty. Through this process, the 2,153 candidate features were distilled down to just 10 radiomic features, which were then combined into a single composite metric called the radiomics score, or rad-score.

In parallel, the investigators examined 18 clinical variables, ranging from demographic characteristics and body mass index to treatment details and nutritional status. Multivariable analysis identified two independent clinical risk factors for moderate-to-severe leukopenia: the score on the Nutritional Risk Screening 2002 tool, a validated questionnaire used to gauge malnutrition risk in hospitalized patients, and receipt of chemotherapy as part of the adjuvant regimen. The NRS-2002 finding aligns with a growing body of evidence linking nutritional reserve to treatment tolerance, while the chemotherapy association reflects the added myelosuppressive burden that cytotoxic drugs impose on top of radiation.

The performance of the resulting models tells a nuanced story. In the training cohort of 106 patients, the rad-score alone achieved an area under the receiver operating characteristic curve, or AUC, of 0.743, while the clinical model reached 0.753. Neither was sufficient on its own. But when the radiomic signature and the clinical variables were merged into a nomogram, a graphical scoring tool that translates multiple inputs into a single probability estimate, the combined model achieved an AUC of 0.830 in the training cohort. In the internal validation cohort of 44 patients, the nomogram maintained an AUC of 0.800 with a 95 percent confidence interval of 0.641 to 0.908, while the clinical model alone dropped to 0.586 and the rad-score to 0.664. The pattern is consistent: image-derived features and clinical factors capture complementary information, and the fusion of the two outperforms either component.

The study’s authors are careful to frame their findings as hypothesis-generating rather than practice-changing. The research was single-center and retrospective, meaning the data came from existing records rather than a prospectively designed trial. The cohort of 150 patients was randomly divided into a training set of 106 and an internal validation set of 44 using a single random split, a design that tests the model on unseen data from the same institution but says nothing about whether the model generalizes to patients at other hospitals with different scanners, protocols, and populations. Radiomic features are notoriously sensitive to variations in CT acquisition parameters, reconstruction kernels, and slice thickness, which is why independent multicenter validation is considered the essential next step before any radiomics nomogram enters routine clinical use.

The combined model’s internal validation result is promising, but it does not establish that using the score improves patient outcomes or safely changes monitoring or treatment. Independent multicenter and prospective evaluation would be needed before relying on the tool in clinical practice. The present study supports further investigation of imaging-based risk estimation, not a change to routine care.

Editorial clarification: Suggestions that this model could justify fewer blood tests or changes in treatment scheduling were not demonstrated by the study and have been removed.

Subject of Research: Radiomics and machine learning for predicting radiotherapy toxicity in cervical cancer

Article Title: CT-based radiomics nomogram for predicting leukopenia in surgically treated stage IB-IIA cervical cancer following (chemo) radiotherapy

Article References: Song, T., Wang, Y., Ding, J., Shou, H., Chen, Y., Zhang, H., & Xu, H. (2026). CT-based radiomics nomogram for predicting leukopenia in surgically treated stage IB-IIA cervical cancer following (chemo) radiotherapy. BMC Cancer. https://doi.org/10.1186/s12885-026-17086-9

Image Credits: AI Generated

DOI: 10.1186/s12885-026-17086-9

Keywords: radiomics, cervical cancer, leukopenia, nomogram, radiotherapy, machine learning, predictive medicine, CT-based, predicting, surgically, treated, stage

The research, conducted by Minying Li, Yan Fang, Yuhong Han, Tuantuan Li, Yan Liu, and Lichang Chen of the Department of Clinical Laboratory at the Second People’s Hospital of Fuyang City in China, enrolled 183 patients with chronic hepatitis B between March 2024 and March 2025. All participants had serum HBV DNA below 10 IU/mL, the detection limit of the quantitative PCR assay used, but HBV RNA at or above 100 copies/mL, the positivity threshold of a capture-probe-based RNA assay. Of these, 87 patients who had developed cirrhosis formed the observation group, while 96 patients without cirrhosis served as controls. All patients were on antiviral therapy, and laboratory measurements were collected after treatment.

The diagnosis of cirrhosis was not based on liver biopsy, which was not routinely performed because of its invasive nature. Instead, the reference standard combined liver stiffness measurement by transient elastography—defined as 12.0 kilopascals or higher—with ultrasonographic features such as a nodular liver surface, blunt liver edge, and splenomegaly, and/or clinical signs of portal hypertension including platelet counts below 100 × 10⁹/L or imaging-confirmed varices. Where biopsy results were available, they were prioritized and showed good concordance with the non-invasive criteria. Patients co-infected with other hepatitis viruses, those with malignancy, autoimmune liver disease, or severe cardiovascular, renal, or hematopoietic disease were excluded to keep the comparison clean.

PNI is a composite score first proposed by Buzby and colleagues in 1980 for perioperative risk assessment. It is calculated as ten times the serum albumin concentration in grams per deciliter plus 0.005 times the total lymphocyte count, and it captures the intertwined status of nutrition and cellular immunity. MLR, by contrast, is simply the monocyte count divided by the lymphocyte count, and it reflects the balance between innate and adaptive immune activity. Both indices can be derived from a standard complete blood count and a routine biochemistry panel, making them far cheaper and more accessible than molecular assays or elastography.

The comparison between groups revealed a coherent biological pattern. Cirrhotic patients had significantly lower white blood cell, lymphocyte, and monocyte counts, along with reduced total protein, albumin, and prealbumin, while aspartate aminotransferase (AST) and gamma-glutamyl transferase (GGT) were significantly higher. Notably, the mean and median values of most of these individual parameters remained within normal reference ranges in both groups, which is precisely why the composite indices mattered. PNI was significantly lower in the cirrhosis group (52.44 versus 57.10), and MLR was significantly higher (0.21 versus 0.16), with both differences highly significant. Importantly, age, sex, ALT, ALP, and alpha-fetoprotein levels did not differ between groups, establishing baseline comparability.

Diagnostic performance was assessed with receiver operating characteristic analysis. PNI achieved an area under the curve of 0.821 with a sensitivity of 95.8 percent at a cut-off of 53.15, while MLR reached an AUC of 0.743 with a sensitivity of 85.4 percent at a cut-off of 0.195. Strikingly, the established fibrosis scores FIB-4 and APRI managed only AUCs of 0.733 and 0.735, respectively, in the same cohort. The authors attribute this shortfall to the fact that FIB-4 and APRI depend heavily on transaminase and platelet dynamics, which tend to normalize when HBV DNA is suppressed, whereas PNI and MLR capture the nutritional-immune axis and systemic inflammation that continue to shift as cirrhosis develops. Within the cirrhosis group, PNI and MLR were also inversely correlated (r = −0.484, P < 0.001), underscoring that the two markers probe related but distinct facets of the same process.

Multivariable logistic regression, with backward stepwise elimination guided by the Akaike Information Criterion and multicollinearity screened using variance inflation factors, confirmed that both indices were independent predictors. Higher PNI behaved as a protective factor, with an odds ratio of 0.694 per unit increase, while elevated MLR was a risk factor with an odds ratio of 3.176. Because MLR showed a markedly right-skewed distribution that destabilized continuous modeling, the team dichotomized it at the optimal ROC-derived cut-off for screening. From these independent predictors the researchers built a nomogram—a graphical scoring tool that translates individual biomarker values into a personalized probability of cirrhosis. The model achieved a concordance index of 0.829, and calibration curves showed good agreement between predicted and observed probabilities, while decision curve analysis indicated positive net clinical benefit.

The mechanistic story the authors propose is subtle and hinges on the difference between HBV DNA and HBV RNA as virological markers. HBV DNA largely reflects active virion production and correlates with overt cytopathic injury, driving the transaminase spikes and platelet changes that FIB-4 and APRI were designed to detect. Persistent serum HBV RNA, particularly pregenomic RNA, instead signals ongoing transcription from the nuclear cccDNA reservoir—the stable minichromosome that nucleos(t)ide analogues cannot eradicate and the fundamental reason hepatitis B is difficult to cure and prone to relapse after treatment withdrawal. In this DNA-suppressed but RNA-positive state, the inflammatory pattern shifts from cytopathic to a chronic, smoldering, immune-mediated process. Viral RNA acts as a pathogen-associated molecular pattern recognized by Toll-like receptors on monocytes and Kupffer cells, sustaining innate activation and relative lymphopenia, while chronic low-grade inflammation suppresses hepatic albumin synthesis as a negative acute-phase reactant. These shifts preferentially perturb the immuno-nutritional axis that PNI and MLR measure.

The study’s limitations are candidly acknowledged. The retrospective, cross-sectional, single-center design means the findings demonstrate association rather than causation or temporal prediction. The specificity of both biomarkers was modest—56.3 percent for PNI and 57.5 percent for MLR—so false positives are a real risk and the markers should be interpreted alongside imaging and other clinical parameters rather than as standalone tests. The nomogram has not yet been externally validated, and incomplete medical records prevented reliable accounting of antiviral regimens, treatment durations, HBeAg status, and HBV RNA levels, leaving room for residual confounding. The authors call for multicenter prospective studies and suggest exploring whether nutritional interventions guided by these biomarkers could improve outcomes.

Even with those caveats, the practical appeal is considerable. Unlike HBV RNA or quantitative HBsAg testing, which require specialized molecular platforms that may be unavailable in resource-limited settings, PNI and MLR can be calculated from laboratory data that virtually every hospital already generates. The Fuyang team’s work suggests that in the growing population of patients whose HBV DNA has been silenced but whose cccDNA quietly persists, the most informative signals of advancing liver disease may lie not in viral load measurements but in the humble complete blood count and albumin value—indices that cost almost nothing and are already sitting in the medical record.

Subject of Research: Non-invasive diagnosis of cirrhosis using PNI and MLR in HBV DNA-negative, HBV RNA-positive chronic hepatitis B patients

Article Title: Diagnostic value of PNI and MLR in cirrhotic patients with negative HBV DNA and positive HBV RNA:A nomogram–based evaluation model

Article References: Li, M., Fang, Y., Han, Y., Li, T., Liu, Y., & Chen, L. (2026). Diagnostic value of PNI and MLR in cirrhotic patients with negative HBV DNA and positive HBV RNA:A nomogram–based evaluation model. Frontiers in Medicine, 13, Article 1919910. https://doi.org/10.3389/fmed.2026.1919910

Image Credits: AI Generated

DOI: 10.3389/fmed.2026.1919910

Keywords: hepatitis B, HBV DNA, HBV RNA, cirrhosis, Prognostic Nutritional Index, monocyte-to-lymphocyte ratio, nomogram, FIB-4, APRI, liver fibrosis, biomarkers, cccDNA

The research team, led by Zhishan Wang and Shujiao Chen at the Third Affiliated People’s Hospital of Fujian University of Traditional Chinese Medicine, enrolled 167 adults with prediabetes and categorized them into four distinct muscle-adipose phenotypes based on their appendicular skeletal muscle index (ASMI) and visceral fat level (VFL). These phenotypes—high-muscle/low-fat, high-muscle/high-fat, low-muscle/high-fat, and low-muscle/low-fat—allowed the investigators to disentangle the independent and combined effects of muscle and fat on insulin resistance, measured using the homeostasis model assessment of insulin resistance (HOMA-IR).

What emerged was a striking metabolic paradox. Participants with the high-muscle/high-fat (HM/HF) phenotype carried the heaviest metabolic burden, displaying the highest levels of fasting blood glucose, triglycerides, and TyG-derived indices. Even more counterintuitively, the appendicular skeletal muscle index was positively correlated with HOMA-IR—defying the expected protective association between greater muscle mass and better insulin sensitivity. This finding suggests that under conditions of energy surplus, skeletal muscle mass may escape net loss and even increase in absolute terms due to heightened mechanical loading, yet myocellular insulin sensitivity and mitochondrial oxidative capacity decline markedly, coinciding with intramuscular fat accumulation.

The study’s hierarchical regression analysis revealed that the HM/HF phenotype was the strongest independent predictor of HOMA-IR, remaining significant after full adjustment for age, sex, and lipid parameters (B = 3.264, P < 0.001). In contrast, the protective effect of the low-muscle/low-fat (LM/LF) phenotype was attenuated to marginal significance after adjusting for blood lipids (B = −0.705, P = 0.082), suggesting that this association largely reflects low visceral fat rather than muscle mass per se. The final model accounted for 36.0% of the variance in HOMA-IR, with body composition phenotypes alone explaining 26.8%.

Perhaps most intriguingly, the researchers found no statistically significant multiplicative interaction between low muscle mass and high visceral fat (P = 0.089), arguing against the hypothesis that these two factors act synergistically to amplify insulin resistance in this population. Instead, the findings point toward visceral fat accumulation as the dominant driver of insulin resistance, with muscle status exerting a secondary or conditional influence within this framework. This aligns with the work of Lingvay and colleagues, who posit that excessive fat accumulation is the fundamental upstream etiology of type 2 diabetes mellitus.

The study also uncovered compelling sex-specific patterns. The HM/HF phenotype showed a particularly strong effect in men (B = 5.662, P < 0.001), whereas the effect was markedly attenuated in women (B = 1.701, P = 0.002). Conversely, the LM/LF phenotype conferred an independent protective effect exclusively in women (B = −0.914, P = 0.025). These discrepancies may stem from the differential programming of body composition distribution and insulin sensitivity by sex hormones. Age also emerged as a potential modifier: among participants younger than 60 years, HM/HF was strongly associated with increased insulin resistance (B = 4.425, P < 0.001), whereas in those aged 60 or older, this association was no longer significant. Instead, triglycerides became the only significant positive predictor in the older group (B = 0.545, P = 0.047), suggesting that in older adults, circulating lipid disturbances may replace body composition patterns as the more dominant factor associated with insulin resistance.

Beyond the muscle-fat phenotyping, the researchers identified two powerful metabolic markers that together form a practical screening tool. The triglyceride-glucose index combined with waist-to-height ratio (TyG-WHtR) emerged as the strongest correlate of HOMA-IR among all TyG-derived indices, and multivariable logistic regression confirmed it as an independent risk factor for insulin resistance with an odds ratio of 8.231 (95% CI: 2.998–22.602, P < 0.001). For each 1-unit increase in TyG-WHtR, the risk of insulin resistance increased approximately 7.2-fold. Meanwhile, high-density lipoprotein cholesterol (HDL-C) served as an independent protective factor (OR = 0.183, 95% CI: 0.044–0.765, P = 0.020), with each 1-unit increase associated with an approximately 81.7% reduction in insulin resistance risk. The Cohen’s d effect sizes for both markers exceeded the conventional large-effect threshold of 0.8, underscoring their clinical significance.

The researchers then constructed a nomogram integrating TyG-WHtR and HDL-C to predict insulin resistance risk in prediabetic individuals. The combined model achieved an area under the receiver operating characteristic curve (AUC) of 0.815 (95% CI: 0.748–0.881), outperforming TyG-WHtR alone (AUC = 0.793) and reverse-coded HDL-C alone (AUC = 0.733). Notably, while TyG-WHtR alone showed high specificity (90.6%) but low sensitivity (49.1%), the combined model achieved a markedly improved sensitivity of 84.2%, though specificity decreased to 47.2%. This profile makes the combined model more suitable for initial screening rather than confirmation: a negative result can help rule out insulin resistance, while a positive result should prompt further diagnostic evaluation.

Model validation was thorough and internally consistent. The Hosmer-Lemeshow test yielded P = 0.650, indicating no evidence of departure from perfect calibration. Calibration curves derived from 1,000 bootstrap resamples closely followed the ideal diagonal line, with the bias-corrected curve nearly coinciding with the ideal line and no clear evidence of overfitting. Decision curve analysis demonstrated that, across threshold probabilities of 0.2 to 0.8, the net benefit of the combined model remained consistently greater than both the treat-all and treat-none strategies. The clinical impact curve showed that the model maintained high event capture with manageable intervention burden up to a cost-benefit ratio of approximately 3:2. These mutually reinforcing results support the combined HDL-C and TyG-WHtR model as a simple and well-calibrated tool for clinical risk stratification in primary care settings.

The study’s conclusions carry important implications for how clinicians and patients think about body composition in the prediabetic state. Rather than relying solely on body mass index or body weight, the findings suggest that integrating visceral fat level and appendicular skeletal muscle index into a combined muscle-adipose phenotype enables a more granular assessment of insulin resistance risk. Maintaining moderate muscle mass in the context of low visceral adiposity is likely to confer the greatest metabolic benefit, while the accumulation of visceral fat—rather than low muscle mass alone—appears to be the primary metabolic threat. The researchers acknowledge several limitations, including the use of bioelectrical impedance analysis rather than gold-standard MRI or CT for body composition assessment, the relatively limited sample size from a single center, and the cross-sectional design that precludes causal inference. Larger multicenter studies with longitudinal follow-up will be needed to validate the phenotype classification and assess the long-term prognostic value of the nomogram model. Nevertheless, this work represents a meaningful step toward personalized metabolic risk assessment, moving the field beyond one-size-fits-all approaches and toward a framework that recognizes the complex interplay between muscle, fat, sex, and age in shaping metabolic health.

Subject of Research: Muscle-adipose phenotypes and insulin resistance in prediabetes

Article Title: Discordant muscle-adipose phenotypes and insulin resistance in prediabetes: sex-specific associations and predictive utility of TyG-WHtR combined with HDL-C

Article References: Wang, Z., Lin, C., Xie, Y., Huang, L., Xiao, S., Chen, H., Ge, H., Zhou, Y., Chen, J., Wu, W., Wang, X., & Chen, S. (2026). Discordant muscle-adipose phenotypes and insulin resistance in prediabetes: sex-specific associations and predictive utility of TyG-WHtR combined with HDL-C. Frontiers in Endocrinology, 17, Article 1937778. https://doi.org/10.3389/fendo.2026.1937778

Image Credits: AI Generated

DOI: 10.3389/fendo.2026.1937778

Keywords: prediabetes, insulin resistance, body composition, visceral fat, skeletal muscle, TyG-WHtR, HDL-C, nomogram, sarcopenic obesity, metabolic syndrome, sex differences, HOMA-IR