How to predict the 10-year mortality risk of patients with localized prostate cancer?

Editor's comment Evaluating the expected survival time of patients with localized prostate cancer is very important for the communication between medical staff and patients.

In this study, researchers developed a tool to assess the 10-year risk of death from other causes in patients with localized prostate cancer.

The researchers said that the tool can help clinicians and patients better communicate to make treatment and follow-up decisions.

Research Introduction Accurately assessing the expected survival time of prostate cancer patients is essential for making treatment decisions.

For patients with low- and medium-risk localized prostate cancer, regardless of the treatment plan, the average risk of death from prostate cancer within 10 years after diagnosis is 5% to 8%, and death due to causes other than prostate cancer (called "other causes") The risk range of is relatively large, ranging from 6% to more than 90%, depending on the patient’s age, health status and social factors.

Therefore, it is very important to determine the patient's risk of death from other causes.

However, the current real-world application of tools for assessing individualized mortality risk has certain limitations.

Moreover, existing tools do not integrate age and comorbidity information with patient reporting factors (such as physical function and general health status), and these factors are increasingly considered to be useful for comprehensive assessment of health status and death from other causes.
Risk has value.

This study sought to determine the pre-treatment patient reporting factors to assess the 10-year risk of death from other causes in patients newly diagnosed with localized prostate cancer.

Research Methods From January 1, 1998 to December 31, 2009, the researchers identified 2,425 prostate cancer patients who were younger than 80 years old and were newly diagnosed with clinical stage T1-T3a from the SEER-MHOS database, and were followed up to February 28, 2013.

Based on age, patient-reported comorbidities, SF-36 health survey composition scores and items, daily activities, and sociodemographic characteristics, the researchers developed a Fine & Gray competitive risk model for the 10-year risk of death from other causes.

The model identification and calibration are compared with the predictions in the mortality risk estimation of the Social Security Survival Sheet.

The results of the study showed that the average age of patients at diagnosis was 73 years old.

The Charls comorbidity index (CCI) score of most (54%) patients is about 0 points (range: 0-7 points), of which 21% of patients are ≥2 points.

Compared with surviving patients, patients who died from other causes were older, had a higher burden of comorbidities, had worse physical health (PCS) scores, had worse overall health, and were more likely to be smokers at the time of diagnosis.
And the proportion of being married is even lower.

Overall, 67% of patients received prostate cancer treatment, most of which were radiation therapy.

It is worth noting that 56% of patients who died from other causes within 10 years after diagnosis received radiotherapy.

In Fine & Gray's competitive risk regression analysis, with a median follow-up of 7.
7 years, 3.
7% of patients died of prostate cancer, and 24.
3% of patients died of other causes.

The median follow-up time for survivors was 8.
9 years.

The pre-treatment factors reported by patients with the highest predictive value for death from other causes include age at diagnosis, approximate CCI value, general health status (good and very good vs.
average and very poor), smoking status at diagnosis, and marital status (married) vs.
other).

According to the predictive value of the variables, age provides the most information, and marital status provides the least information.

The final model includes a model with more information about age, CCI approximation, general health status, smoking status at diagnosis, marital status, physical and mental health, and daily activities.

In a model that includes age and CCI approximations, other socioeconomic factors (education level, family income, house ownership) and ethnic factors have no significant correlation with the risk of death from other causes, but they belong to the “other” category, which is not clinically feasible.
Explain the definition and represent a very small number of patients.

Except for poor general health and a CCI score of approximately 7 points (<11 people), no significant interaction was observed, so these two items were excluded from the final model.

In the sensitivity analysis including patients over 80 years of age, the validity of the model also remained similar.

Table: The median 10-year risk of death from other causes predicted by the final model risk score of the 10-year risk of death from other causes is 20% (interquartile range 15%-29%).

The model is well calibrated in the SEER-MHOS population, and the prediction rate of death from other causes is observed as the risk increases; in contrast, in this study, the social insurance life table overestimates the risk of most patients.

The dynamic risk calculation tool can be obtained from
 Conclusion This study provides a predictive tool that uses a simple set of patient report characteristics that can better identify patients with high and low risk of non-prostate cancer death in 10 years compared with the social insurance life table.

References Daniel M.
FrendlID, Gordon FitzGerald, Mara M.
Epstein, et al.
Predicting the 10-year risk of death from other causes in men with localized prostate cancer using patient-reported factors :Development of a tool[J].
PLoS ONE 15(12): e0240039.