Statistical Interaction in the Survival Analysis of Early Breast Cancer using Registry Data: Role of Breast Conserving Surgery and Radiotherapy

The effect of adjuvant radiotherapy on mortality differs according to primary tumor location in women with node-positive breast cancer

PURPOSE

To evaluate the prognostic significance of primary tumor location and to examine whether the effect of adjuvant radiotherapy on survival varies according to tumor location in women with axillary node-positive (ALN+) breast cancer (BC).

PATIENTS AND METHODS

Data were abstracted from the SEER database for 24,410 women aged 25-95 years, diagnosed between 1988-1997 with nonmetastatic T1-T2, ALN+ BC. Subgroup analyses were performed using interactions within proportional hazards models. Event was defined as death from any cause. Prognostic variables were selected using Akaike Information Criteria. Joint significances of subgroups were evaluated with Wald test.

RESULTS

Median follow-up was 10 years. In joint models, statistically significant interactions were found between tumor location, nodal involvement, type of surgery, and radiotherapy. Factorial presentation of interactions showed consistent 13% proportional reduction of mortality in all subgroups, except in women with medial tumors with > or = 4 ALN+ treated with mastectomy. In this subgroup, use of radiotherapy was associated with a 16% proportional increase in mortality.

CONCLUSION

Medial tumor location is a significant adverse prognostic factor that should be considered in treatment decision- making for women with ALN+ BC. Improved survival was observed with radiotherapy use in all subgroups, except in women with medial tumors with > or = 4 ALN+ treated with postmastectomy radiotherapy. These findings raise concern that the favorable effect of radiotherapy may be offset by excess toxicities in the latter subgroup.

No nodal cutoff in node-positive breast cancer women treated with mastectomy

PURPOSE

To search for a prognostic cutoff in the number of involved axillary lymph-nodes, in T1-T2 node-positive (N+) breast cancer treated with mastectomy without radiotherapy.

MATERIALS AND METHODS

Women aged 51-64 years presenting with a first primary T1-T2, N+, M0 unilateral carcinoma diagnosed in 1988-97 were selected from the SEER 9-registries. Cutoffs from 1 to 23 nodes were scanned with Cox models that included diagnostic year, age, area, race, marital status, histology, grade, hormone receptors, tumor location, size, and nodes examined. Outcomes were breast cancer specific and any-cause death.

RESULTS

The overall mortality hazard ratio of a higher number of involved nodes as compared with a lower number ranged from 1.78 [95% confidence 1.58-1.99] with 1-node cutoff, to 3.65 [2.52-5.29] with 23-nodes cutoff. Breast cancer specific mortality hazard ratio ranged from 1.99 [1.73-2.29] with 1-node cutoff, to 4.63 [3.11-6.91] with 23-nodes cutoff. The graph of the hazard ratios showed a continuously increasing risk of overall and breast cancer specific mortality, without any identifiable cutpoint change.

CONCLUSION

The results show no prognostic cutoff in the number of involved axillary lymph nodes in patients who did not receive radiotherapy. The rationale of nodal cutoff as an indicator for post-mastectomy adjuvant treatment in node-positive patients should be questioned.