Development and validation of an inflammatory biomarkers model to predict gastric cancer prognosis: a multi-center cohort study in China.

BACKGROUND: Inflammatory factors have increasingly become a more cost-effective prognostic indicator for gastric cancer (GC). The goal of this study was to develop a prognostic score system for gastric cancer patients based on inflammatory indicators. METHODS: Patients' baseline characteristics and anthropometric measures were used as predictors, and independently screened by multiple machine learning(ML) algorithms. We constructed risk scores to predict overall survival in the training cohort and tested risk scores in the validation. The predictors selected by the model were used in multivariate Cox regression analysis and developed a nomogram to predict the individual survival of GC patients. RESULTS: A 13-variable adaptive boost machine (ADA) model mainly comprising tumor stage and inflammation indices was selected in a wide variety of machine learning models. The ADA model performed well in predicting survival in the validation set (AUC = 0.751; 95% CI: 0.698, 0.803). Patients in the study were split into two sets - "high-risk" and "low-risk" based on 0.42, the cut-off value of the risk score. We plotted the survival curves using Kaplan-Meier analysis. CONCLUSION: The proposed model performed well in predicting the prognosis of GC patients and could help clinicians apply management strategies for better prognostic outcomes for patients.

Cadonilimab with chemotherapy in HER2-negative gastric or gastroesophageal junction adenocarcinoma: the phase 1b/2 COMPASSION-04 trial.

Treatment with anti-programmed cell death protein 1 (PD-1) therapy and chemotherapy prolongs the survival of patients with unresectable advanced or metastatic gastric or gastroesophageal junction (GEJ) adenocarcinoma. The benefit from anti-PD-1 therapy is enriched in patients with programmed cell death 1 ligand 1 (PD-L1) combined positive score (CPS)-positive or CPS-high tumors compared with patients with PD-L1 CPS-negative or CPS-low tumors. In this phase 1b/2 study, we evaluated the efficacy and safety of cadonilimab, a bispecific antibody targeting PD-1 and cytotoxic T-lymphocyte antigen-4, plus chemotherapy as first-line treatment in patients with human epidermal growth factor receptor 2-negative unresectable advanced or metastatic gastric or GEJ adenocarcinoma. The primary endpoint was the recommended phase 2 dose (RP2D) for phase 1b and the objective response rate for phase 2. Secondary endpoints included disease control rate, duration of response, time to response, progression-free survival, overall survival (OS) and safety. The primary endpoint was met. No dose-limiting toxicities were observed during dose escalation in phase 1b; the recommended phase 2 dose was determined as 6 mg kg-1 every 2 weeks. The objective response rate was 52.1% (95% confidence interval (CI) = 41.6-62.5), consisting of complete and partial responses in 4.3% and 47.9% of patients, respectively. The median duration of response, progression-free survival and OS were 13.73 months (95% CI = 7.79-19.12), 8.18 months (95% CI = 6.67-10.48) and 17.48 months (95% CI = 12.35-26.55), respectively. The median OS in patients with a PD-L1 CPS ≥ 5 was 20.32 months (95% CI = 4.67-not estimable); in patients with a PD-L1 CPS < 1, the median OS reached 17.64 months (95% CI = 11.63-31.70). The most common treatment-related grade 3 or higher adverse events were decreased neutrophil count (19.1%), decreased platelet count (16.0%), anemia (12.8%) and decreased leukocyte count (8.5%). No new safety signal was identified. The current regimen showed promising clinical activity and manageable safety in patients with gastric or GEJ adenocarcinoma regardless of PD-L1 expression. Chinadrugtrials.org.cn registration: CTR20182027.

Identification of prognostic signatures in remnant gastric cancer through an interpretable risk model based on machine learning: a multicenter cohort study.

OBJECTIVE: The purpose of this study was to develop an individual survival prediction model based on multiple machine learning (ML) algorithms to predict survival probability for remnant gastric cancer (RGC). METHODS: Clinicopathologic data of 286 patients with RGC undergoing operation (radical resection and palliative resection) from a multi-institution database were enrolled and analyzed retrospectively. These individuals were split into training (80%) and test cohort (20%) by using random allocation. Nine commonly used ML methods were employed to construct survival prediction models. Algorithm performance was estimated by analyzing accuracy, precision, recall, F1-score, area under the receiver operating characteristic curve (AUC), confusion matrices, five-fold cross-validation, decision curve analysis (DCA), and calibration curve. The best model was selected through appropriate verification and validation and was suitably explained by the SHapley Additive exPlanations (SHAP) approach. RESULTS: Compared with the traditional methods, the RGC survival prediction models employing ML exhibited good performance. Except for the decision tree model, all other models performed well, with a mean ROC AUC above 0.7. The DCA findings suggest that the developed models have the potential to enhance clinical decision-making processes, thereby improving patient outcomes. The calibration curve reveals that all models except the decision tree model displayed commendable predictive performance. Through CatBoost-based modeling and SHAP analysis, the five-year survival probability is significantly influenced by several factors: the lymph node ratio (LNR), T stage, tumor size, resection margins, perineural invasion, and distant metastasis. CONCLUSIONS: This study established predictive models for survival probability at five years in RGC patients based on ML algorithms which showed high accuracy and applicative value.

The potential of machine learning models to identify malnutrition diagnosed by GLIM combined with NRS-2002 in colorectal cancer patients without weight loss information.

BACKGROUND & AIMS: The key step of the Global Leadership Initiative on Malnutrition (GLIM) is nutritional risk screening, while the most appropriate screening tool for colorectal cancer (CRC) patients is yet unknown. The GLIM diagnosis relies on weight loss information, and bias or even failure to recall patients' historical weight can cause misestimates of malnutrition. We aimed to compare the suitability of several screening tools in GLIM diagnosis, and establish machine learning (ML) models to predict malnutrition in CRC patients without weight loss information. METHODS: This multicenter cohort study enrolled 4487 CRC patients. The capability of GLIM diagnoses combined with four screening tools in predicting survival probability was compared by Kaplan-Meier curves, and the most accurate one was selected as the malnutrition reference standard. Participants were randomly assigned to a training cohort (n = 3365) and a validation cohort (n = 1122). Several ML approaches were adopted to establish models for predicting malnutrition without weight loss data. We estimated feature importance and reserved the top 30% of variables for retraining simplified models. The area under the receiver operating characteristic curve (AUC), accuracy, sensitivity, and specificity were calculated to assess and compare model performance. RESULTS: NRS-2002 was the most suitable screening tool for GLIM diagnosis in CRC patients, with the highest hazard ratio (1.59; 95% CI, 1.43-1.77). A total of 2076 (46.3%) patients were malnourished diagnosed by GLIM combined with NRS-2002. The simplified random forest (RF) model outperformed other models with an AUC of 0.830 (95% CI, 0.805-0.854), and accuracy, sensitivity and specificity were 0.775, 0.835 and 0.742, respectively. We deployed an online application based on the simplified RF model to accurately estimate malnutrition probability in CRC patients without weight loss information (https://zzuwtt1998.shinyapps.io/dynnomapp/). CONCLUSIONS: Nutrition Risk Screening 2002 was the optimal initial nutritional risk screening tool in the GLIM process. The RF model outperformed other models, and an online prediction tool was developed to properly identify patients at high risk of malnutrition.

Exploring the optimal indicator of short-term peridiagnosis weight dynamics to predict cancer survival: A multicentre cohort study.

BACKGROUND: Body weight and its changes have been associated with cancer outcomes. However, the associations of short-term peridiagnosis weight dynamics in standardized, clinically operational time frames with cancer survival remain largely unknown. This study aimed to screen for and evaluate the optimal indicator of short-term peridiagnosis weight dynamics to predict overall survival (OS) in patients with cancer. METHODS: This multicentre cohort study prospectively collected data from 7460 patients pathologically diagnosed with cancer between 2013 and 2019. Body weight data were recorded 1 month before, at the time of and 1 month following diagnosis. By permuting different types (point value in kg, point height-adjusted value in kg/m2, absolute change in kg or relative change in percentage) and time frames (prediagnosis, postdiagnosis or peridiagnosis), we generated 12 different weight-related indicators and compared their prognostic performance using Harrell's C-index, integrated discrimination improvement, continuous net reclassification improvement and time-dependent C-index. We analysed associations of peridiagnosis relative weight change (RWC) with OS using restricted cubic spine (RCS), Kaplan-Meier analysis and multivariable-adjusted Cox regression models. RESULTS: The study enrolled 5012 males and 2448 females, with a median age of 59 years. During a median follow-up of 37 months, 1026 deaths occurred. Peridiagnosis (1 month before diagnosis to 1 month following diagnosis) RWC showed higher prognostic performance (Harrell's C-index = 0.601, 95% confidence interval [CI] = [0.583, 0.619]) than other types of indicators including body mass index (BMI), absolute weight change, absolute BMI change, prediagnosis RWC and postdiagnosis RWC in the study population (all P < 0.05). Time-dependent C-index analysis also indicated that peridiagnosis RWC was optimal for predicting OS. The multivariable-adjusted RCS analysis revealed an N-shaped non-linear association between peridiagnosis RWC and OS (PRWC < 0.001, Pnon-linear < 0.001). Univariate survival analysis showed that the peridiagnosis RWC groups could represent distinct mortality risk stratifications (P < 0.001). Multivariable survival analysis showed that, compared with the maintenance group (weight change < 5%), the significant (gain >10%, hazard ratio [HR] = 0.530, 95% CI = [0.413, 0.680]) and moderate (gain 5-10%, HR = 0.588, 95% CI = [0.422, 0.819]) weight gain groups were both associated with improved OS. In contrast, the moderate (loss 5-10%, HR = 1.219, 95% CI = [1.029, 1.443]) and significant (loss >10%, HR = 1.280, 95% CI = [1.095, 1.497]) weight loss groups were both associated with poorer OS. CONCLUSIONS: The prognostic performance of peridiagnosis RWC is superior to other weight-related indicators in patients with cancer. The findings underscore the importance of expanding the surveillance of body weight from at diagnosis to both past and future, and conducting it within clinically operational time frames, in order to identify and intervene with patients who are at risk of weight change-related premature deaths.

Integrated neutrophil-to-lymphocyte ratio and handgrip strength better predict survival in patients with cancer cachexia.

OBJECTIVES: Systemic inflammation and skeletal muscle strength play crucial roles in the development and progression of cancer cachexia. In this study we aimed to evaluate the combined prognostic value of neutrophil-to-lymphocyte ratio (NLR) and handgrip strength (HGS) for survival in patients with cancer cachexia. METHODS: This multicenter cohort study involved 1826 patients with cancer cachexia. The NLR-HGS (NH) index was defined as the ratio of neutrophil-to-lymphocyte ratio to handgrip strength. Harrell's C index and receiver operating characteristic (ROC) curve analysis were used to assess the prognosis of NH. Kaplan-Meier analysis and Cox regression models were used to evaluate the association of NH with all-cause mortality. RESULTS: Based on the optimal stratification, 380 women (NH > 0.14) and 249 men (NH > 0.19) were classified as having high NH. NH has shown greater predictive value compared to other indicators in predicting the survival of patients with cancer cachexia according to the 1-, 3-, and 5-y ROC analysis and Harrell's C index calculation. Multivariate survival analysis showed that higher NH was independently associated with an increased risk of death (hazard ratio = 1.654, 95% confidence interval = 1.389-1.969). CONCLUSION: This study demonstrates that the NH index, in combination with NLR and HGS, is an effective predictor of the prognosis of patients with cancer cachexia. It can offer effective prognosis stratification and guidance for their treatment.

Value of the modified Patient-Generated Subjective Global Assessment in indicating the need for nutrition intervention and predicting overall survival in patients with malignant tumors in at least two organs.

BACKGROUND: Although the Patient-Generated Subjective Global Assessment (PG-SGA) is a reference standard used to assess a patient's nutrition status, it is cumbersome to administer. The aim of the present study was to estimate the value of a simpler and easier-to-use modified PG-SGA (mPG-SGA) to evaluate the nutrition status and need for intervention in patients with malignant tumors present in at least two organs. METHODS: A total of 591 patients (343 male and 248 female) were included from the INSCOC study. A Pearson correlation analysis was conducted to assess the correlation between the mPG-SGA and nutrition-related factors, with the optimal cut-off defined by a receiver operating characteristic curve (ROC). The consistency between the mPG-SGA and PG-SGA was compared in a concordance analysis. A survival analysis was used to determine the effects of nutritional intervention among different nutrition status groups. Univariable and multivariable Cox analyses were applied to evaluate the association of the mPG-SGA with the all-cause mortality. RESULTS: The mPG-SGA showed a negative association with nutrition-related factors. Individuals with an mPG-SGA ≥ 5 (rounded from 4.5) were considered to need nutritional intervention. Among the malnourished patients (mPG-SGA ≥ 5), the overall survival (OS) of those who received nutrition intervention was significantly higher than that of patients who did not. However, the OS was not significantly different in the better-nourished patients (mPG-SGA < 5). CONCLUSION: Our findings support that the mPG-SGA is a feasible tool that can be used to guide nutritional interventions and predict the survival of patients with malignant tumors affecting at least two organs.

Efficacy, Safety, and Population Pharmacokinetics of MW032 Compared With Denosumab for Solid Tumor-Related Bone Metastases: A Randomized, Double-Blind, Phase 3 Equivalence Trial.

Importance: The bioequivalence of denosumab biosimilar has yet to be studied in a 53-week, multicenter, large-scale, and head-to-head trial. A clinically effective biosimilar may help increase access to denosumab in patients with solid tumor-related bone metastases. Objectives: To establish the biosimilarity of MW032 to denosumab in patients with solid tumor-related bone metastases based on a large-scale head-to-head study. Design, Setting, and Participants: In this 53-week, randomized, double-blind, phase 3 equivalence trial, patients with solid tumors with bone metastasis were recruited from 46 clinical sites in China. Overall, 856 patients were screened and 708 eligible patients were randomly allocated to receive either MW032 or denosumab. Interventions: Patients were randomly assigned (1:1) to receive MW032 or reference denosumab subcutaneously every 4 weeks until week 49. Main Outcomes and Measures: The primary end point was percentage change from baseline to week 13 of natural logarithmic transformed urinary N-telopeptide/creatinine ratio (uNTx/uCr). Results: Among the 701 evaluable patients (350 in the MW032 group and 351 in the denosumab group), the mean (range) age was 56.1 (22.0-86.0) years and 460 patients were women (65.6%). The mean change of uNTx/uCr from baseline to week 13 was -72.0% (95% CI, -73.5% to -70.4%) in the MW032 group and -72.7% (95% CI, -74.2% to -71.2%) in the denosumab group. These percent changes corresponded to mean logarithmic ratios of -1.27 and -1.30, or a difference of 0.02. The 90% CI for the difference (-0.04 to 0.09) was within the equivalence margin (-0.13 to 0.13); the mean changes of uNTx/uCr and bone-specific alkaline phosphatase (s-BALP) at each time point were also similar during 53 weeks. The differences of uNTx/uCr change were 0.015 (95% CI, -0.06 to 0.09), -0.02 (95% CI, -0.09 to 0.06), -0.05 (95% CI, -0.13 to 0.03) and 0.001 (95% CI, -0.10 to 0.10) at weeks 5, 25, 37, and 53, respectively. The differences of s-BALP change were -0.006 (95% CI, 0.06 to 0.05), 0.00 (95% CI, -0.07 to 0.07), -0.085 (95% CI, -0.18 to 0.01), -0.09 (95% CI, -0.20 to 0.02), and -0.13 (95% CI, -0.27 to 0.004) at weeks 5, 13, 25, 37 and 53, respectively. No significant differences were observed in the incidence of skeletal-related events (-1.4%; 95% CI, -5.8% to 3.0%) or time to first on-study skeletal-related events (unadjusted HR, 0.86; P = .53; multiplicity adjusted HR, 0.87; P = .55) in the 2 groups. Conclusions and Relevance: MW032 and denosumab were biosimilar in efficacy, population pharmacokinetics, and safety profile. Availability of denosumab biosimilars may broaden the access to denosumab and reduce the drug burden for patients with advanced tumors. Trial Registration: ClinicalTrials.gov Identifier: NCT04812509.

Association of possible sarcopenia with all-cause mortality in patients with solid cancer: A nationwide multicenter cohort study.

OBJECTIVES: The concept of possible sarcopenia (PS) was recently introduced to enable timely intervention in settings without the technologies required to make a full diagnosis of sarcopenia. This study aimed to investigate the association between PS and all-cause mortality in patients with solid cancer. DESIGN: Retrospective observational study. SETTING AND PARTICIPANTS: 13,736 patients with 16 types of solid cancer who were ≥18 years old. MEASUREMENTS: The presence of both a low calf circumference (men <34 cm or women <33 cm) and low handgrip strength (men <28 kg or women <18 kg) was considered to indicate PS. Harrell's C-index was used to assess prognostic value and the association of PS with mortality was estimated by calculating multivariable-adjusted hazard ratios (HRs). RESULTS: The study enrolled 7207 men and 6529 women (median age = 57.8 years). During a median follow-up of 43 months, 3150 deaths occurred. PS showed higher Harrell's C-index (0.549, 95%CI = [0.541, 0.557]) than the low calf circumference (0.541, 95%CI = [0.531, 0.551], P = 0.037) or low handgrip strength (0.542, 95%CI = [0.532, 0.552], P = 0.026). PS was associated with increased mortality risk in both univariate (HR = 1.587, 95%CI = [1.476, 1.708]) and multivariable-adjusted models (HR = 1.190, 95%CI = [1.094, 1.293]). Sensitivity analyses showed that the association of PS with mortality was robust in different covariate subgroups, which also held after excluding those patients who died within the first 3 months (HR = 1.162, 95%CI = [1.060, 1.273]), 6 months (HR = 1.150, 95%CI = [1.039, 1.274]) and 12 months (HR = 1.139, 95%CI = [1.002, 1.296]) after enrollment. CONCLUSION: PS could independently and robustly predict all-cause mortality in patients with solid cancer. These findings imply the importance of including PS assessment in routine cancer care to provide significant prognostic information to help mitigate sarcopenia-related premature deaths.

Discovery of distinct cancer cachexia phenotypes using an unsupervised machine-learning algorithm.

OBJECTIVES: Cancer cachexia is a debilitating condition with widespread negative effects. The heterogeneity of clinical features within patients with cancer cachexia is unclear. The identification and prognostic analysis of diverse phenotypes of cancer cachexia may help develop individualized interventions to improve outcomes for vulnerable populations. The aim of this study was to show that the machine learning-based cancer cachexia classification model generalized well on the external validation cohort. METHODS: This was a nationwide multicenter observational study conducted from October 2012 to April 2021 in China. Unsupervised consensus clustering analysis was applied based on demographic, anthropometric, nutritional, oncological, and quality-of-life data. Key characteristics of each cluster were identified using the standardized mean difference. We used logistic and Cox regression analysis to evaluate 1-, 3-, 5-y, and overall mortality. RESULTS: A consensus clustering algorithm was performed for 4329 patients with cancer cachexia in the discovery cohort, and four clusters with distinct phenotypes were uncovered. From clusters 1 to 4, the clinical characteristics of patients showed a transition from almost unimpaired to mildly, moderately, and severely impaired. Consistently, an increase in mortality from clusters 1 to 4 was observed. The overall mortality rate was 32%, 40%, 54%, and 68%, and the median overall survival time was 21.9, 18, 16.7, and 13.6 mo for patients in clusters 1 to 4, respectively. Our machine learning-based model performed better in predicting mortality than the traditional model. External validation confirmed the above results. CONCLUSIONS: Machine learning is valuable for phenotype classifications of patients with cancer cachexia. Detection of clinically distinct clusters among cachexic patients assists in scheduling personalized treatment strategies and in patient selection for clinical trials.