gms | German Medical Science

Background: Breast cancer is the most frequent malignancy among the female population. The survival rates of breast cancer patients have been increasing considerably during the past decades. Regular follow-up after breast cancer treatment aims for an early detection of locoregional breast cancer recurrences (LRR) to improve the patients’ outcome. By estimating individual’s 5-year recurrence-risks based on different patient- tumor- and treatment characteristics, the Dutch INFLUENCE-nomogram can assist health professionals and patients in developing personalized risk-based follow-up pathways. This prognostic nomogram is based on over 37,000 patients of the Netherlands cancer registry (NCR) from the years 2003 to 2006.

Research question: Until today it is unclear whether the nomogram is generalizable to foreign populations and health care systems, which would contribute to demonstrate its clinical relevance. The objective of this study is to externally validate the prediction tool on non-Dutch patients with additional emphasis on important patient subgroups.

Methods: Data for this external validation derive from a large clinical cancer registry in southern Germany, covering a population of 1.1 million. Patients with curative resection of early-stage breast cancer, diagnosed between 2000 and 2012, were included in the analysis. To account for selection bias, a sensitivity analysis comparing LRR rates of included patients and patients excluded due to missing data was performed. For every included patient, an individual LRR-risk was estimated by the INFLUENCE-nomogram. Its predictive ability was tested by comparing estimated and observed LRR-probabilities using the Hosmer–Lemeshow goodness-of-fit test and C-statistic based on the receiver-operator-characteristic (ROC) curve.

Results: Finally, 6520 patients fulfilling all inclusion criteria without missing data in any nomogram variable were included. In this German validation-cohort, 2.8% of the patients developed an LRR within 5 years after primary surgery (n = 184). The LRR-rate among the excluded patients was 2.9%, which, according to the sensitivity analysis, is not significantly different from the included patients’ LRR-rate (p = 0.902). While the INFLUENCE-nomogram generally underestimates the actual LRR-risk of the German patients (p < 0.001), its discriminative ability is comparable to the one observed in the original Dutch modeling-cohort (C-statistic German validation-cohort: 0.73, CI 0.69–0.77 vs. C-statistic Dutch modeling-cohort: 0.71, CI 0.69–0.73). Similar results were obtained in most of the subgroup analyses stratified by age, type of surgery and intrinsic biological subtypes.

Discussion: The present study is the first one testing the Dutch INFLUENCE-nomogram with external data from another country, which is an essential step towards its implementation in the daily clinical practice. A reason for the INFLUENCE-nomogram underestimating the German LRR-risk that might be that the LRR-rate in the German cohort is slightly – but not significantly – higher. Moreover, it could reflect moderate differences in therapy perception between the two populations. For clinical use, accuracy is less important than discriminative ability, anyway. To develop personalized follow-up pathways, physicians most probably will use the INFLUENCE-nomogram together with some kind of cut-off. The ROC-curve depicts sensitivity and specificity for every possible threshold which can be used with the INFLUENCE-tool. The C-statistic, therefore, represents the discriminative ability of the algorithm. For the 5-year overall LRR-risk algorithm, the C-statistic was 0.71 in the Dutch modelling-cohort; almost the same value was obtained by the first external validation with another Dutch cohort from 2007 and 2008. With the German patients analyzed within this study, the C-statistic was even slightly larger (0.73); this indicates good external validity.

Practical implication: This study is a good example for beneficial cooperation between international cancer registries. Its outcomes underline the generalizability of the recently developed INFLUENCE-nomogram beyond the Dutch population. The model performance of INFLUENCE could be enhanced in future by incorporating additional risk factors for LRR.

Funding: Funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – project number 417891978.