Survival prediction in mesothelioma using a scalable Lasso regression model: instructions for use and initial performance using clinical predictors

Our plain-English summary of Survival prediction in mesothelioma using a scalable Lasso regression model: instructions for use and initial performance using clinical predictors, by A.C. Kidd, M. McGettrick, S. Tsim, et al (BMJ Open Respiratory Research, 2018).

The study

• Authors: A.C. Kidd, M. McGettrick, S. Tsim, D.L. Halligan, M. Bylesjo, K.G. Blyth
• Journal: BMJ Open Respiratory Research
• Published: 2018;5(1):e000240
• Cited by: 161 articles, as counted by NIH iCite
• Read it: PubMed · Publisher · Free full text

Abstract

Introduction: Accurate prognostication is difficult in malignant pleural mesothelioma (MPM). We developed a set of robust computational models to quantify the prognostic value of routinely available clinical data, which form the basis of published MPM prognostic models.

Methods: Data regarding 269 patients with MPM were allocated to balanced training (n=169) and validation sets (n=100). Prognostic signatures (minimal length best performing multivariate trained models) were generated by least absolute shrinkage and selection operator regression for overall survival (OS), OS <6 months and OS <12 months. OS prediction was quantified using Somers DXY statistic, which varies from 0 to 1, with increasing concordance between observed and predicted outcomes. 6-month survival and 12-month survival were described by area under the curve (AUC) scores.

Results: Median OS was 270 (IQR 140-450) days. The primary OS model assigned high weights to four predictors: age, performance status, white cell count and serum albumin, and after cross-validation performed significantly better than would be expected by chance (mean DXY0.332 (±0.019)). However, validation set DXY was only 0.221 (0.0935-0.346), equating to a 22% improvement in survival prediction than would be expected by chance. The 6-month and 12-month OS signatures included the same four predictors, in addition to epithelioid histology plus platelets and epithelioid histology plus C-reactive protein (mean AUC 0.758 (±0.022) and 0.737 (±0.012), respectively). The <6-month OS model demonstrated 74% sensitivity and 68% specificity. The <12-month OS model demonstrated 63% sensitivity and 79% specificity. Model content and performance were generally comparable with previous studies.

Conclusions: The prognostic value of the basic clinical information contained in these, and previously published models, is fundamentally of limited value in accurately predicting MPM prognosis. The methods described are suitable for expansion using emerging predictors, including tumour genomics and volumetric staging.

In plain English

This study used hospital records of 269 patients with malignant pleural mesothelioma to test how well routine clinical information predicts survival. Asbestos exposure was not studied. The patients were split into a training group of 169 and a validation group of 100. Of 20 clinical measures collected, serum lactate dehydrogenase was dropped for missing data, and a statistical method called Lasso regression built models from the remaining 19. Median survival was 270 days. The main survival model relied mostly on age, performance status (a rating of daily functioning), white cell count and serum albumin. In the validation group its concordance score was 0.221 (0.0935 to 0.346), which the authors described as a 22% improvement over chance. Models predicting death within 6 months and within 12 months added epithelioid tumour type plus platelets or C-reactive protein. The 6-month model had 74% sensitivity and 68% specificity, and the 12-month model had 63% sensitivity and 79% specificity. The authors concluded that basic clinical information is of limited value for accurately predicting mesothelioma prognosis. They listed retrospective data collection, missing data and validation only within their own patient group as limitations.

We wrote this plain-English summary from the full article with the help of AI, and a second AI model checked each sentence against the article. It is not the authors' wording; the abstract and the article are their own account.

What the study did and found

• Design: Retrospective prognostic modelling study using Lasso regression, with patients split into a training set and a reserved validation set
• Who was studied: 269 patients with malignant pleural mesothelioma diagnosed in the West of Scotland Cancer Network between January 2008 and April 2014, identified from a hospital pathology archive in Glasgow, UK
• Exposure: Asbestos exposure was not studied. The study collected 20 routinely available clinical predictor variables, including age, sex, histological subtype, performance status, deprivation, comorbidity, blood and inflammatory markers, and aspirin use. Serum LDH was dropped because of missing data, so the models were built from 19.
• Main results:
  • Median overall survival in the 269 patients was 270 days (IQR 140 to 450).
  • The main overall survival model gave high weight to four predictors: age, performance status, white cell count and serum albumin.
  • Within the training set (cross-validation), the main survival model had a mean DXY of 0.332 (plus or minus 0.019).
  • In the validation set, the main survival model had a DXY of 0.221 (0.0935 to 0.346), which the authors described as a 22% improvement over chance.
  • The 6-month and 12-month models used the same four predictors plus epithelioid histology, with platelets added in the 6-month model and C-reactive protein added in the 12-month model.
  • The 6-month model's AUC in the validation set was 0.74 (0.638 to 0.836).
  • The 12-month model's AUC in the validation set was 0.794 (0.688 to 0.883).
  • The model predicting death within 6 months had 74% sensitivity and 68% specificity, with 70% accuracy.
  • The model predicting death within 12 months had 63% sensitivity and 79% specificity, with 69% accuracy.
  • The authors did not identify deprivation as a major prognostic factor in this study.
  • The authors found no evidence of a clinically important prognostic effect of aspirin use.
  • 67 of the 269 patients (24.9%) received chemotherapy during follow-up.
• The authors' conclusion: The authors concluded that the basic clinical information in their models, and in previously published models, is of limited value for accurately predicting prognosis in malignant pleural mesothelioma.
• Limitations the authors state:
  • Data were partly collected retrospectively, introducing potential recall and omission bias.
  • Cases came from a pathology archive, so frail patients who never had a tissue diagnosis were not included.
  • Validation used an internal group of patients only; external validation is still needed.
  • Many cases had missing data for some variables.
  • Performance status was recorded inconsistently, so a best estimate was used.

Topics

Medical terms in this study, explained in our Lexicon: Mesothelioma, Pleural Mesothelioma, Overall Survival, Performance Status, Epithelioid, Histology, Prognosis, Staging.

Related studies

• Survival in Good Performance Malignant Pleural Mesothelioma Patients; Prognostic Factors and Predictors of Response — M. Rahouma, H. Aziz, et al. (2017)
• A Novel Clinical Prediction Model for Prognosis in Malignant Pleural Mesothelioma Using Decision Tree Analysis — F.J. Brims, T.M. Meniawy, et al. (2016)
• Prediction modelling using routine clinical parameters to stratify survival in Malignant Pleural Mesothelioma patients undergoing cytoreductive surgery — E.J.A. Harris, S. Kao, et al. (2019)
• Prediction modeling using routine clinical parameters to stratify survival in malignant pleural mesothelioma patients complicated with malignant pleural effusion — S. Zhang, Y. Zhang, et al. (2021)
• Assessment of potential predictors of calretinin and mesothelin to improve the diagnostic performance to detect malignant mesothelioma: results from a population-based cohort study — S. Casjens, D.G. Weber, et al. (2017)
• Ordinal regression models for epidemiologic data — B.G. Armstrong, M. Sloan (1989)

Browse more research in BMJ Open Respiratory Research, or return to medical research.