Survival prediction score: a simple but age-dependent method predicting prognosis in patients undergoing palliative radiotherapy

AutoScore-Survival: Developing interpretable machine learning-based time-to-event scores with right-censored survival data

BACKGROUND

Scoring systems are highly interpretable and widely used to evaluate time-to-event outcomes in healthcare research. However, existing time-to-event scores are predominantly created ad-hoc using a few manually selected variables based on clinician's knowledge, suggesting an unmet need for a robust and efficient generic score-generating method.

METHODS

AutoScore was previously developed as an interpretable machine learning score generator, integrating both machine learning and point-based scores in the strong discriminability and accessibility. We have further extended it to the time-to-event outcomes and developed AutoScore-Survival, for generating time-to-event scores with right-censored survival data. Random survival forest provided an efficient solution for selecting variables, and Cox regression was used for score weighting. We implemented our proposed method as an R package. We illustrated our method in a study of 90-day survival prediction for patients in intensive care units and compared its performance with other survival models, the random survival forest, and two traditional clinical scores.

RESULTS

The AutoScore-Survival-derived scoring system was more parsimonious than survival models built using traditional variable selection methods (e.g., penalized likelihood approach and stepwise variable selection), and its performance was comparable to survival models using the same set of variables. Although AutoScore-Survival achieved a comparable integrated area under the curve of 0.782 (95% CI: 0.767-0.794), the integer-valued time-to-event scores generated are favorable in clinical applications because they are easier to compute and interpret.

CONCLUSIONS

Our proposed AutoScore-Survival provides a robust and easy-to-use machine learning-based clinical score generator to studies of time-to-event outcomes. It gives a systematic guideline to facilitate the future development of time-to-event scores for clinical applications.

The LabBM score is an excellent survival prediction tool in patients undergoing palliative radiotherapy

Background and aim

The prognostic assessment of patients referred for palliative radiotherapy can be conducted by site-specific scores. A quick assessment that would cover the whole spectrum could simplify the working day of clinicians who are not specialists for a particular disease site. This study evaluated a promising score, the LabBM (validated for brain metastases), in patients treated for other indications.

Materials and methods

The LabBM score was calculated in 375 patients by assigning 1 point each for C-reactive protein and lactate dehydrogenase above the upper limit of normal, and 0.5 points each for hemoglobin, platelets and albumin below the lower limit of normal. Uni- and multivariate analyses were performed.

Results

Median overall survival gradually decreased with increasing point sum (range 25.1-1.1 months). When grouped according to the original three-tiered model, excellent discrimination was found. Patients with 0-1 points had a median survival of 15.7 months. Those with 1.5-2 points had a median survival of 5.8 months. Finally, those with 2.5-3.5 points had a median survival of 3.2 months (all p-values ≤ 0.001).

Conclusion

The LabBM score, which is derived from inexpensive blood tests and easy to use, stratified patients into three very distinct prognostic groups and deserves further validation.

Patient-reported symptoms before palliative radiotherapy predict survival differences

BACKGROUND

Widely used prognostic scores, e. g., for brain or bone metastases, are based on disease- and patient-related factors such as extent of metastases, age and performance status, which were available in the databases used to develop the scores. Few groups were able to include patient-reported symptoms. In our department, all patients were assessed with the Edmonton Symptom Assessment System (ESAS, a one-sheet questionnaire addressing 11 major symptoms and wellbeing on a numeric scale of 0-10) at the time of treatment planning since 2012. Therefore, we analyzed the prognostic impact of baseline ESAS symptom severity.

METHODS

Retrospective review of 102 patients treated with palliative radiotherapy (PRT) between 2012 and 2015. All ESAS items were dichotomized (below/above median). Uni- and multivariate analyses were performed to identify prognostic factors for survival.

RESULTS

The most common tumor types were prostate, breast and non-small cell lung cancer, predominantly with distant metastases. Median survival was 6 months. Multivariate analysis resulted in six significant prognostic factors. These were ESAS pain while not moving (median 3), ESAS appetite (median 5), Eastern Cooperative Oncology Group (ECOG) performance status, pleural effusion/metastases, intravenous antibiotics at start or within 2 weeks before PRT and no systemic cancer treatment.

CONCLUSIONS

Stronger pain while not moving and reduced appetite (below/above median) predicted significantly shorter survival. Development of new prognostic scores should include patient-reported symptoms and other innovative parameters because they were more important than primary tumor type, age and other traditional baseline parameters.

Overall survival after reirradiation of spinal metastases - independent validation of predictive models

BACKGROUND

It is unknown if survival prediction tools (SPTs) sufficiently predict survival in patients who undergo palliative reirradiation of spinal metastases. We therefore set out to clarify if SPTs can predict survival in this patient population.

METHODS

We retrospectively analyzed spinal reirradiations performed (n = 58, 52 patients, 44 included in analysis). SPTs for patients with spinal metastases were identified and compared to a general palliative score and to a dedicated SPT to estimate prognosis in palliative reirradiation independent of site (SPT-Nieder).

RESULTS

Consistently in all tests, SPT-Nieder showed best predictive performance as compared to other tools. Items associated with survival were general condition (KPS), liver metastases, and steroid use. Other factors like primary tumor site, pleural effusion, and bone metastases were not correlated with survival. We adapted an own score to the data which performed comparable to SPT-Nieder but avoids the pleural effusion item. Both scores showed good performance in identifying long-term survivors with late recurrences.

CONCLUSIONS

Survival prediction in case of spinal reirradiation is possible with sufficient predictive separation. Applying SPTs in case of reirradiation helps to identify patients with good life expectancy who might benefit from dose escalation or longer treatment courses.

Clinical and genetic factors related to cancer-induced bone pain and bone pain relief

OBJECTIVE

The study objective was to evaluate whether there are clinical or genetic differences between patients with cancer-induced bone pain (CIBP) and patients with non-CIBP, and, in the CIBP group, in those with good versus poor opioid response.

MATERIALS AND METHODS

A total of 2,294 adult patients with cancer who were receiving opioids for moderate or severe pain were included in the European Pharmacogenetic Opioid Study. Pain intensity and pain relief were measured using the Brief Pain Inventory. Linkage disequilibrium of 112 single nucleotide polymorphisms was evaluated in 25 candidate genes, and 43 haplotypes were assessed. Correlations among demographical factors, disease-related factors, genetic factors, CIBP, and pain relief were analyzed by logistic regression models corrected for multiple testing. Patients with bone metastases and bone/soft tissue pain were defined as having prevalent bone pain (CIBP population). This population was compared with patients who had other types of cancer pain (non-CIBP).

RESULTS

A total of 577 patients (26.2%) had CIBP, and 1,624 patients (73.8%) had non-CIBP. Patients with CIBP had more breakthrough cancer pain episodes (64.2% vs. 56.4%, p = .001), had significantly higher pain interference in "walking ability in the past 24 hours" (p < .0001), used more adjuvant drugs (84.1% vs. 78.3%, p = .003), and had a higher, albeit nonsignificant, median overall survival (3.8 vs. 2.9 months, p = .716) than patients with non-CIBP. None of the examined haplotypes exceeded p values corrected for multiple testing for the investigated outcomes.

CONCLUSION

Patients with CIBP who were taking opioids had a clinical profile slightly different from that of the non-CIBP group. However, no specific genetic pattern emerged for CIBP versus non-CIBP or for responsive versus nonresponsive patients with CIBP.