Why we should take care of the competing risk bias in survival analysis: A phase II trial on the toxicity profile of radiotherapy for prostate cancer

Clinical Outcomes and Relevance of Composite V12 Gy in Patients With Four or More Brain Metastases Treated With Single Fraction Stereotactic Radiosurgery

AIMS

Tissue V12Gy (total brain volume receiving 12Gy including target) can predict for late toxicity in single target benign disease treated with stereotactic radiosurgery (SRS). The value of this metric remains uncertain for multiple brain metastases. This retrospective cohort study reports the outcomes and evaluates the predictors of toxicity in patients with four or more brain metastases treated with single-fraction SRS.

MATERIALS AND METHODS

Two hundred twenty-six patients with 2160 metastases treated from 2014-21 were retrospectively studied. Symptomatic late toxicity (new/progressive neurological symptoms ≥3 months post SRS) with magnetic resonance imaging (MRI) changes suggestive of treatment effect were analysed. Kaplan-Meier and competing risk analysis was used to assess survival and toxicity respectively.

RESULTS

median number of metastases/patient was 6 (range: 4-41) and median composite tissue V12Gy (inclusive of planning target volume (PTV)) was 11.3 cc (IQR: 6.1 cc-17.1 cc). Sixteen out of the 226 patients developed symptomatic late radiation adverse event (R-AE), and the cumulative incidence was 4.9% at 1 year and 6.9% at 2 years. The total target volume was significantly predictive of the risk of late R-AE. Volume of the largest lesion, V12Gy and V15Gy did not predict for late R-AE, but plotted graphs showed suggestions of linear relationships between dosimetric parameters and late R-AE.

CONCLUSION

Within the limitations of this study, the cumulative incidence of symptomatic toxicity remains acceptable despite routinely accepting a composite tissue V12Gy in excess of 10 cc to treat multiple brain metastases.

ADVANCES IN KNOWLEDGE

V12Gy has limitations as a plan quality metric in multiple brain metastases treated with SRS. There is insufficient evidence to have a defined target limit as <10 cc.

The hepatocellular carcinoma risk in patients with HBV-related cirrhosis: a competing risk nomogram based on a 4-year retrospective cohort study

Objective

The study aimed to build and validate a competitive risk nomogram to predict the cumulative incidence of hepatocellular carcinoma (HCC) for patients with hepatitis B virus (HBV)-related cirrhosis.

Methods

A total of 1401 HBV-related cirrhosis patients were retrospectively enrolled from January 1, 2011 to December 31, 2014. Application of 20 times imputation dealt with missing data using multiple imputation by chained equations (MICE). The patients were randomly divided into a training set (n = 1017) and a validation set (n = 384) at a ratio of 3:1. A prediction study was carried out using a competing risk model, where the event of interest was HCC and the competing events were death and liver transplantation, and subdistribution hazard ratios (sHRs) with 95% CIs were reported. The multivariate competing risk model was constructed and validated.

Results

There was a negligible difference between the original database and the 20 imputed datasets. At the end of follow-up, the median follow-up time was 69.9 months (interquartile range: 43.8-86.6). There were 31.5% (442/1401) of the patients who developed HCC, with a 5-year cumulative incidence of 22.9 (95%CI, 20.8%-25.2%). The univariate and multivariate competing risk regression and construction of the nomogram were performed in 20 imputed training datasets. Age, sex, antiviral therapy history, hepatitis B e antigen, alcohol drinking history, and alpha-fetoprotein levels were included in the nomogram. The area under receiver operating characteristic curve values at 12, 24, 36, 60, and 96 months were 0.68, 0.69, 0.70, 0.68, and 0.80, and the Brier scores were 0.30, 0.25, 0.23, 0.21, and 0.20 in the validation set. According to the cumulative incidence function, the nomogram effectively screened out high-risk HCC patients from low-risk patients in the presence of competing events (Fine-Gray test p < 0.001).

Conclusion

The competitive risk nomogram was allowed to be used for predicting HCC risk in individual patients with liver cirrhosis, taking into account both the association between risk factors and HCC and the modifying effect of competition events on this association.

Risk of developing psoriatic arthritis in psoriasis cohorts with arthralgia: exploring the subclinical psoriatic arthritis stage

OBJECTIVE

Subjects with subclinical psoriatic arthritis (PsA), defined as the presence of arthralgia in psoriasis (PsO), are at higher risk of PsA but scant real-world data exist. Our aims were to (1) estimate the probability of PsA development in subclinical PsA, (2) characterise subclinical PsA symptoms and (3) determine the clinical patterns at PsA diagnosis.

METHODS

Patients with PsO, mainly subclinical PsA, were evaluated longitudinally in two European cohorts. The key outcome was new-onset PsA. Musculoskeletal symptoms including inflammatory and non-inflammatory symptoms before PsA diagnosis were collected. Occurrence of PsA was analysed with survival analysis and cumulative incidence functions (CIFs).

RESULTS

384 patients with PsO were included with a mean follow-up of 33.0 (±20.9) months. 311 of 384 (80.9%) had subclinical PsA with a PsA incidence rate of 7.7 per 100 patient-years. Subclinical PsA displayed a higher risk of PsA development compared with PsO (HR=11.7 (95% CI 1.57 to 86.7), p=0.016). The probability of new-onset PsA estimated by the CIF was 9.4% (95% CI 4.7% to 10.6%) at month 12 and 22.7% (95% CI 17.2% to 28.6%) at month 36. 58.9% of cases reported inflammatory symptoms in the months immediately prior to PsA diagnosis but prior non-inflammatory symptoms were evident in 83.9% prior to PsA diagnosis. Peripheral joint swelling was the predominant PsA presentation pattern (82.1%).

CONCLUSIONS

The probability of PsA development among subclinical PsA was relatively high, emphasising the importance of emergent musculoskeletal symptoms when aiming for PsA prevention. Joint swelling was the dominant feature in new-onset PsA, likely reflecting clinical confidence in recognising joint swelling.

Communicating and understanding statistical measures when quantifying the between-group difference in competing risks

Competing risks issues are common in clinical trials and epidemiological studies for patients in follow-up who may experience a variety of possible outcomes. Under such competing risks, two hazard-based statistical methods, cause-specific hazard (CSH) and subdistribution hazard (SDH), are frequently used to assess treatment effects among groups. However, the outcomes of the CSH-based and SDH-based methods have a close connection with the proportional hazards (CSH or SDH) assumption and may have an non-intuitive interpretation. Recently, restricted mean time lost (RMTL) has been used as an alternative summary measure for analysing competing risks, due to its clinical interpretability and robustness to the proportional hazards assumption. Considering the above approaches, we summarize the differences between hazard-based and RMTL-based methods from the aspects of practical interpretation, proportional hazards model assumption and the selection of restricted time points, and propose corresponding suggestions for the analysis of between-group differences under competing risks. Moreover, an R package 'cRMTL' and corresponding step-by-step guidance are available to help users for applying these approaches.

Statistical models versus machine learning for competing risks: development and validation of prognostic models

BACKGROUND

In health research, several chronic diseases are susceptible to competing risks (CRs). Initially, statistical models (SM) were developed to estimate the cumulative incidence of an event in the presence of CRs. As recently there is a growing interest in applying machine learning (ML) for clinical prediction, these techniques have also been extended to model CRs but literature is limited. Here, our aim is to investigate the potential role of ML versus SM for CRs within non-complex data (small/medium sample size, low dimensional setting).

METHODS

A dataset with 3826 retrospectively collected patients with extremity soft-tissue sarcoma (eSTS) and nine predictors is used to evaluate model-predictive performance in terms of discrimination and calibration. Two SM (cause-specific Cox, Fine-Gray) and three ML techniques are compared for CRs in a simple clinical setting. ML models include an original partial logistic artificial neural network for CRs (PLANNCR original), a PLANNCR with novel specifications in terms of architecture (PLANNCR extended), and a random survival forest for CRs (RSFCR). The clinical endpoint is the time in years between surgery and disease progression (event of interest) or death (competing event). Time points of interest are 2, 5, and 10 years.

RESULTS

Based on the original eSTS data, 100 bootstrapped training datasets are drawn. Performance of the final models is assessed on validation data (left out samples) by employing as measures the Brier score and the Area Under the Curve (AUC) with CRs. Miscalibration (absolute accuracy error) is also estimated. Results show that the ML models are able to reach a comparable performance versus the SM at 2, 5, and 10 years regarding both Brier score and AUC (95% confidence intervals overlapped). However, the SM are frequently better calibrated.

CONCLUSIONS

Overall, ML techniques are less practical as they require substantial implementation time (data preprocessing, hyperparameter tuning, computational intensity), whereas regression methods can perform well without the additional workload of model training. As such, for non-complex real life survival data, these techniques should only be applied complementary to SM as exploratory tools of model's performance. More attention to model calibration is urgently needed.

Risk factors for severe lower extremity ischemia following venoarterial extracorporeal membrane oxygenation: an analysis using a nationwide inpatient database

Objectives

Venoarterial extracorporeal membrane oxygenation is increasingly being used as a life-saving modality in critically ill patients. Despite its necessity, severe lower extremity ischemia associated with venoarterial extracorporeal membrane oxygenation remains a potentially devastating complication. We aimed to investigate the incidence and risk factors for severe lower extremity ischemia requiring fasciotomy or amputation following venoarterial extracorporeal membrane oxygenation.

Methods

All patients who received venoarterial extracorporeal membrane oxygenation during hospitalization were identified in a Japanese national inpatient database from July 1, 2010 to March 31, 2018. The primary outcome was occurrence of severe lower extremity ischemia that required fasciotomy or amputation. We used cause-specific proportional hazard models to examine the associations between potential risk factors and outcomes. We also performed a competing-risk analysis to estimate the cause-specific HR for severe lower extremity ischemia using a multivariable competing-risk Cox proportional hazard model with adjustment for potential risk factors.

Results

A total of 29 231 patients who underwent venoarterial extracorporeal membrane oxygenation during hospitalization were identified. Of these, 98 patients (0.3%) had lower extremity ischemia requiring fasciotomy or amputation. The young group (≤18 years) had a significantly higher proportion of severe lower extremity ischemia cases than the adult (19-59 years) and elderly (≥60 years) groups (1.4%, 0.5%, and 0.2%, respectively; p<0.001). In a multivariable competing-risk Cox proportional hazards regression model, younger age (HR 3.06; 95% CI 1.33 to 7.02; p<0.008) and consciousness disturbance on admission (HR 2.53; 95% CI 1.60 to 3.99; p<0.001) were significantly associated with higher likelihood of severe lower extremity ischemia.

Conclusion

In this study using a nationwide database, younger age and consciousness disturbance on admission were associated with higher risk of severe lower extremity ischemia following venoarterial extracorporeal membrane oxygenation.

Level of evidence

Level Ⅲ-prognostic and epidemiological.

A retrospective study of clinicopathologic and molecular features of inoperable early-stage non-small cell lung cancer treated with stereotactic ablative radiotherapy

BACKGROUND/PURPOSE

Stereotactic ablative radiotherapy (SABR) is the treatment of choice for medically inoperable, early-stage non-small cell lung cancer (ES-NSCLC). The influence of oncogenic driver alterations and comorbidities are not well known. Here we present treatment outcomes based on clinicopathologic features and molecular profiles.

METHODS

We retrospectively analyzed patients treated with SABR for inoperable ES-NSCLC. Molecular features of oncogenic driver alterations included EGFR, ALK, and ROS1. Comorbidities were assessed using the age-adjusted Charlson Comorbidity Index (ACCI). Survival was calculated using the Kaplan-Meier method. The Cox regression model was performed for univariate and multivariate analyses of prognostic factors. Competing risk analysis was used to evaluate the cumulative incidence of disease progression.

RESULTS

From 2008 to 2020, 100 patients (median age: 82 years) were enrolled. The majority of patients were male (64%), ever-smokers (60%), and had adenocarcinoma (65%). With a median follow-up of 21.5 months, the median overall survival (OS) and real-world progression-free survival were 37.7 and 25.1 months, respectively. The competing-risk-adjusted 3-year cumulative incidences of local, regional, and disseminated failure were 8.2%, 14.5%, and 31.2%, respectively. An ACCI ≥7 was independently associated with inferior OS (hazard ratio [HR] 2.45, p = 0.03). Tumor size ≥4 cm (HR 4.16, p < 0.001) was the most important independent prognostic factor predicting real-world progression. EGFR mutation status had no impact on the outcomes.

CONCLUSION

SABR provides excellent local control in ES-NSCLC, although disseminated failures remains a major concern. ACCI is the best indicator for OS, while tumor sizes ≥4 cm predicts poor disease control.