# Adaption of the Memorial Sloan Kettering Cancer Center Nomograms for the Prediction of Prostate Cancer–specific Death in Sweden: A Population-based Cohort Study

## TL;DR

This study adapts and validates cancer risk prediction tools for Swedish prostate cancer patients, showing they work well in predicting cancer-specific deaths.

## Contribution

The study adapts and validates the MSKCC nomograms for prostate cancer-specific mortality in a Swedish population.

## Key findings

- The preoperative and postoperative models showed high discrimination for prostate cancer-specific death (C-index: 0.81 and 0.87).
- Comorbidities added minimal value to predicting prostate cancer-specific mortality.
- Models were well calibrated and suitable for clinical use in Sweden.

## Abstract

The pre- and postoperative Memorial Sloan Kettering Cancer Center nomograms updated to the Swedish setting show high discrimination for prostate cancer–specific death at 10 yr. The models were well calibrated and can be used in clinical practice in Sweden.

Prognostication is a cornerstone of the clinical management of prostate cancer. This study aims to update the pre- and postoperative Memorial Sloan Kettering Cancer Center (MSKCC) nomograms for the prediction of 10-yr prostate cancer–specific mortality in the competing risk setting in Sweden, and to evaluate the added value of comorbidities.

A cohort study was conducted including all men in the National Prostate Cancer Register of Sweden diagnosed with localised prostate cancer in 2007–2020, who underwent radical prostatectomy. Follow-up was until December 31, 2022. We used cause-specific Cox proportional hazard models to obtain the cumulative incidence of prostate cancer–specific and other-cause mortality. The models were validated in terms of discrimination (concordance [C] index) and calibration by internal-external validation in six Swedish health care regions and by bootstrapping (N = 500).

The cohort included 31 106 men, of whom 629 died from prostate cancer and 2415 died from other causes during a median follow-up of 8.3 yr (interquartile range: 5.2, 11.8). Comorbidities added more value to the other-cause mortality model than to the prostate cancer–specific mortality model, and were included in all models. Both the preoperative and the postoperative model showed high discrimination for prostate cancer–specific death (optimism-corrected C-index: 0.81 and 0.87, respectively), but not for other-cause mortality (0.67, both models). All models were well calibrated, with minimal overestimation at the higher range of predicted cumulative incidences for the preoperative, but not for the postoperative, model.

The updated MSKCC nomograms performed well in terms of discrimination and calibration, and can be used in clinical practice in Sweden. In this study, comorbidity added minimal prognostic value for predicting prostate cancer–specific mortality. External validation is advised for application in other populations.

Prognostication is a cornerstone in the clinical management of prostate cancer. In this study, we adapted the best preforming risk classification system, the pre- and postoperative Memorial Sloan Kettering Cancer Center nomograms, for the prediction of prostate cancer–specific death in Swedish setting. The adapted models perform well and can be applied directly to Swedish men with prostate cancer.