Dataset Information

Explainable machine learning model for predicting skeletal muscle loss during surgery and adjuvant chemotherapy in ovarian cancer.

ABSTRACT:

Background

Skeletal muscle loss during treatment is associated with poor survival outcomes in patients with ovarian cancer. Although changes in muscle mass can be assessed on computed tomography (CT) scans, this labour-intensive process can impair its utility in clinical practice. This study aimed to develop a machine learning (ML) model to predict muscle loss based on clinical data and to interpret the ML model by applying SHapley Additive exPlanations (SHAP) method.

Methods

This study included the data of 617 patients with ovarian cancer who underwent primary debulking surgery and platinum-based chemotherapy at a tertiary centre between 2010 and 2019. The cohort data were split into training and test sets based on the treatment time. External validation was performed using 140 patients from a different tertiary centre. The skeletal muscle index (SMI) was measured from pre- and post-treatment CT scans, and a decrease in SMI ≥ 5% was defined as muscle loss. We evaluated five ML models to predict muscle loss, and their performance was determined using the area under the receiver operating characteristic curve (AUC) and F1 score. The features for analysis included demographic and disease-specific characteristics and relative changes in body mass index (BMI), albumin, neutrophil-to-lymphocyte ratio (NLR), and platelet-to-lymphocyte ratio (PLR). The SHAP method was applied to determine the importance of the features and interpret the ML models.

Results

The median (inter-quartile range) age of the cohort was 52 (46-59) years. After treatment, 204 patients (33.1%) experienced muscle loss in the training and test datasets, while 44 (31.4%) patients experienced muscle loss in the external validation dataset. Among the five evaluated ML models, the random forest model achieved the highest AUC (0.856, 95% confidence interval: 0.854-0.859) and F1 score (0.726, 95% confidence interval: 0.722-0.730). In the external validation, the random forest model outperformed all ML models with an AUC of 0.874 and an F1 score of 0.741. The results of the SHAP method showed that the albumin change, BMI change, malignant ascites, NLR change, and PLR change were the most important factors in muscle loss. At the patient level, SHAP force plots demonstrated insightful interpretation of our random forest model to predict muscle loss.

Conclusions

Explainable ML model was developed using clinical data to identify patients experiencing muscle loss after treatment and provide information of feature contribution. Using the SHAP method, clinicians may better understand the contributors to muscle loss and target interventions to counteract muscle loss.

SUBMITTER: Hsu WH

PROVIDER: S-EPMC10570082 | biostudies-literature | 2023 Oct

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Explainable machine learning model for predicting skeletal muscle loss during surgery and adjuvant chemotherapy in ovarian cancer.

Hsu Wen-Han WH Ko Ai-Tung AT Weng Chia-Sui CS Chang Chih-Long CL Jan Ya-Ting YT Lin Jhen-Bin JB Chien Hung-Ju HJ Lin Wan-Chun WC Sun Fang-Ju FJ Wu Kun-Pin KP Lee Jie J

Journal of cachexia, sarcopenia and muscle 20230712 5

<h4>Background</h4>Skeletal muscle loss during treatment is associated with poor survival outcomes in patients with ovarian cancer. Although changes in muscle mass can be assessed on computed tomography (CT) scans, this labour-intensive process can impair its utility in clinical practice. This study aimed to develop a machine learning (ML) model to predict muscle loss based on clinical data and to interpret the ML model by applying SHapley Additive exPlanations (SHAP) method.<h4>Methods</h4>This ...[more]

PMID: 37435785

Similar Datasets

Project description:Purpose: Despite advances in radical surgery and chemotherapy delivery, ovarian cancer is the most lethal gynecologic malignancy. Most of these patients are treated with platinum-based chemotherapies, but there is no biomarker model to guide their responses to these therapeutic agents. We have developed and independently tested our novel multivariate molecular predictors for forecasting patients' responses to individual drugs on a cohort of 58 ovarian cancer patients. Experimental Design: We adapted and applied the previously-published COXEN algorithm to develop molecular predictors for therapeutic responses of patients' tumors based on expression signatures derived from the NCI-60 in vitro drug activities and genomic expression data. Genome-wide candidate biomarkers were first triaged by examining expression patterns of frozen and formalin-fixed paraffin embedded (FFPE) tissue samples. We then identify initial drug sensitivity biomarkers for carboplatin and paclitaxel, respectively. These biomarkers were further narrowed by examining concordant expression patterns between cell lines and a historical set of ovarian cancer patients. Multivariate predictors were obtained from the NCI-60 cell lines and refined using historical patient cohorts. To independent validate these molecular predictors, we performed genome-wide profiling on FFPE samples of 58 ovarian cancer patients obtained prior to adjuvant chemotherapy. Results: Carboplatin predictor significantly stratified platinum sensitive and resistant patients (p = 0.019) with sensitivity = 93%, specificity = 33%, PPV = 65%, and NPV = 78%. Paclitaxel predictor also significantly stratified patients' responses (p = 0.033) with sensitivity = 96%, specificity = 26%, PPV = 61%, and NPV = 86%. The combination predictor for platinum-taxane combination demonstrated a significant survival difference between the predicted responders and nonresponders with median survival of 12.9 months vs. 8.1 months (p = 0.045). Conclusions: COXEN predictors successfully stratified platinum resistance and taxane response in this retrospective cohort, especially based on their FFPE tumor samples. Accurate prediction of chemotherapeutic response, especially to platinum agents is highly clinically relevant and could alter primary management of ovarian cancer. Gene expression data from 58 stage III-IV ovarian cancer patients treated with Carboplatin and Taxol agents

Project description:BackgroundSarcopenia is commonly observed in patients with advanced-stage epithelial ovarian cancer (EOC). However, the effect of body composition changes-during primary debulking surgery (PDS) and adjuvant platinum-based chemotherapy-on outcomes of patients with advanced-stage EOC is unknown. This study aimed to evaluate the association between body composition changes and outcomes of patients with stage III EOC treated with PDS and adjuvant platinum-based chemotherapy.MethodsPre-treatment and post-treatment computed tomography (CT) images of 139 patients with stage III EOC were analysed. All CT images were contrast-enhanced scans and were acquired according to a standardized protocol. The skeletal muscle index (SMI), skeletal muscle radiodensity (SMD), and total adipose tissue index were measured using CT images obtained at the L3 vertebral level. Predictors of overall survival were identified using Cox regression models.ResultsThe median follow-up was 37.9 months. The median duration between pre-treatment and post-treatment CT was 182 days (interquartile range: 161-225 days). Patients experienced an average SMI loss of 1.8%/180 days (95% confidence interval: -3.1 to -0.4; P = 0.01) and SMD loss of 1.7%/180 days (95% confidence interval: -3.3 to -0.03; P = 0.046). SMI and SMD changes were weakly correlated with body mass index changes (Spearman ρ for SMI, 0.15, P = 0.07; ρ for SMD, 0.02, P = 0.82). The modified Glasgow prognostic score was associated with SMI loss (odds ratio: 2.42, 95% confidence interval: 1.03-5.69; P = 0.04). The median time to disease recurrence was significantly shorter in patients with SMI loss ≥5% after treatment than in those with SMI loss <5% or gain (5.4 vs. 11.2 months, P = 0.01). Pre-treatment SMI (1 cm2 /m2 decrease; hazard ratio: 1.08, 95% confidence interval: 1.03-1.11; P = 0.002) and SMI change (1%/180 days decrease; hazard ratio: 1.04, 95% confidence interval: 1.01-1.08; P = 0.002) were independently associated with poorer overall survival. SMD, body mass index, and total adipose tissue index at baseline and changes were not associated with overall survival.ConclusionsSkeletal muscle index decreased significantly during treatment and was independently associated with poor overall survival in patients with stage III EOC treated with PDS and adjuvant platinum-based chemotherapy. The modified Glasgow prognostic score might be a predictor of SMI loss during treatment.

Project description:ObjectiveWe studied the effects of loss of ovarian function (ovariectomy) on muscle mass of gastrocnemius and the mRNA levels of IGF-1, atrogin-1, MuRF-1, and myostatin in an experimental model of rheumatoid arthritis in rats.MethodsWe randomly allocated 24 female Wistar rats (9 weeks, 195.3 ± 17.4 grams) into four groups: control (CT-Sham; n = 6); rheumatoid arthritis (RA; n = 6); ovariectomy without rheumatoid arthritis (OV; n = 6); ovariectomy with rheumatoid arthritis (RAOV; n = 6). We performed the ovariectomy (OV and RAOV) or Sham (CT-Sham or RA) procedures at the same time, fifteen days before the rheumatoid arthritis induction. The RA and RAOV groups were immunized and then were injected with Met-BSA in the tibiotarsal joint. After 15 days of intra-articular injections the animals were euthanized. We evaluated the external manifestations of rheumatoid arthritis (perimeter joint) as well as animal weight, and food intake throughout the study. We also analyzed the cross-sectional areas (CSA) of gastrocnemius muscle fibers in 200 fibers (H&E method). In the gastrocnemius muscle, we analyzed mRNA expression by quantitative real time PCR followed by the Livak method (ΔΔCT).ResultsThe rheumatoid arthritis induced reduction in CSA of gastrocnemius muscle fibers. The RAOV group showed a lower CSA of gastrocnemius muscle fibers compared to RA and CT-Sham groups. Skeletal muscle IGF-1 mRNA increased in arthritics and ovariectomized rats. The increased IGF-1 mRNA was higher in OV groups than in the RA and RAOV groups. Antrogin-1 mRNA also increased in the gastrocnemius muscle of arthritic and ovariectomized rats. However, the increased atrogin-1 mRNA was higher in RAOV groups than in the RA and OV groups. Gastrocnemius muscle MuRF-1 mRNA increased in the OV and RAOV groups, but not in the RA and Sham groups. However, the RAOV group showed higher MuRF-1 mRNA than the OV group. The myostatin gene expression was similar in all groups.ConclusionLoss of ovarian function results in increased loss of skeletal muscle-related ubiquitin ligases atrogin-1 and MuRF-1 in arthritic rats.

Dataset Information

Explainable machine learning model for predicting skeletal muscle loss during surgery and adjuvant chemotherapy in ovarian cancer.

Background

Methods

Results

Conclusions

Publications

Explainable machine learning model for predicting skeletal muscle loss during surgery and adjuvant chemotherapy in ovarian cancer.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets