The impact of local/regional radiotherapy (RT), and its timing, on survival following lumpectomy/mastectomy and systemic chemotherapy in patients with ≥10 positive axillary nodes: Analysis of CALGB 9082

Abstract

In CALGB 9082, patients with breast cancer involving ≥10 axillary nodes without evidence of distant metastasis received 4 cycles of induction CAF chemo and were then randomized to receive either high dose cyclophosphamide, cisplatin, BCNU with autologous marrow/stem-cell support or intermediate-doses of the same drugs. All patients were prescribed to receive local-regional RT following chemo. We herein assess whether there was an association of overall survival with (1) receipt of RT and (2) the timeliness of RT among irradiated patients. This trial randomized 785 patients from 5/91–5/98. Of these, RT information was available for 747. The protocol-specified RT included 50 Gy to the chest wall, supraclavicular nodes in 1.8 Gy daily fractions using conventional techniques. IMN RT was encouraged but not required. Axillary RT was discouraged but permitted. We used proportional hazards models to assess two study questions. First, the overall survival in patients who received (+RT) or did not receive (−RT) were compared. Since patients were not randomized to ±RT, this comparison may be biased. For example, RT was often omitted due to chemo-associated morbidity that may bias against the −RT group. To address this, we used a landmark analysis in which a fixed time after a patient’s study-entry was selected as a landmark time. Patients who died prior to the landmark time were excluded from the analysis in an effort to remove the confounding effect of chemo-associated mortality/morbidity that may have impacted on the decision to use or not use RT. Patients were considered to have had RT only if they received it before the landmark. The dependent variable was survival measured from the landmark time until death due to any cause. We used landmark times of 9, 12, 18 and 24 months post study entry. Question 2: Among +RT patients, the survival in those who started RT within 45 days (the median) from the end of chemo, vs. >45 days were compared. The dependent variable was survival measured from 45 days after the end of chemo. All proportional hazards models controlled for treatment arm (high vs intermediate dose chemo), type of primary surgery (mastectomy vs other), age (<50 vs ≥50) and ER status (neg vs pos). The median follow-up was 8 yrs. For the 747 patients, median age was 44 yrs; 80% Caucasian; 65% stage II and 35% stage III; 69% receptor positive; 77% mastectomy. 91% of the 747 patients were irradiated. Of those who did not receive RT, 73% were on the HD-CPB arm. The survival hazard favored the +RT group at the 9 and 12 mo landmark analyses; hazard ratios [HR] were 1.36 (p = 0.098) and 1.28 (p = 0.31), respectively (the −RT group had an excess hazard of dying that was 36% or 28% higher than the +RT group). HRs favored the −RT group in the 18- and 24-month landmark analyses; HRs were 1.01 (p = 0.97) and 1.05 (p = 0.87), respectively. A test of the association of timeliness of RT with survival had p = 0.090. Patients whose RT started beyond 45 days after chemo had a hazard of dying that was 24% higher than those who started RT within 45 days of completing chemo (HR = 1.24, early RT better). In this comparison of non-randomized groups of patients, there was no definitive/consistent impact of RT, or its timeliness, on survival. Some of the p-values approached statistical significance (p = 0.09–0.10) suggesting that the use of RT, and its delivery within 45 days of chemo, may improve survival. The inability to demonstrate a statistically-significant association between RT usage/timing and survival may be due to 1) biases in selection/timing for RT that are not addressed by a landmark technique, 2) the relatively small number of patients, 3) competing risks of systemic spread in these patients with multiple positive axillary nodes, and 4) the lack of any real effect. Additional analyses of patterns of failure are underway and ill be presented