The study designs of the randomised trials were reported earlier2,7-9
and are summarised in . The first randomised study started in June 1973, and by September 1975, 386 eligible patients with node positive breast cancer had been allocated to receive either no further treatment after radical mastectomy or adjuvant CMF for 12 cycles.2,7
The second randomised trial started in September 1975, and initially premenopausal and postmenopausal women were allocated to receive either 12 or six cycles of postoperative CMF.8
In November 1976, however, random enrolment of postmenopausal women was discontinued, and the study was limited to premenopausal patients; the recruitment was closed in May 1978, and the sample included 324 eligible, node positive, premenopausal women. An observational study between May 1978 and October 1980 included 220 premenopausal patients with node positive breast cancer, all scheduled to receive 12 cycles of CMF. The fourth trial, between December 1980 and October 1985, included 90 women with node negative and oestrogen receptor negative tumours, who were randomised to either no systemic treatment after surgery or 12 cycles of intravenous CMF given every three weeks.9
CMF studies carried out at the Istituto Nazionale Tumori in Milan
The study populations consisted of patients admitted to the Istituto Nazionale Tumori in Milan, Italy. All women who had had surgery (radical mastectomy or conservative surgery and full axillary clearance) for unilateral breast cancer were considered for inclusion in the studies if they had histological evidence that one or more axillary nodes were affected (first three studies) or if they had histologically negative axillary nodes and oestrogen receptor negative tumours (third study).
Patients with locally advanced or metastatic disease, those with a history of previous cancer, and those with concomitant severe non-malignant systemic disease were not eligible.
With the exception of the randomised trial investigating oestrogen receptor negative tumours (the third trial), assessment of the hormone receptors was not mandatory and was done only retrospectively, by using the dextran coated charcoal technique. This was done in all but the first study, in which they were assessed on immunohistochemistry. According to Italian rules at the time, all patients had to give verbal informed consent before being enrolled into each of the studies.
In patients with node positive breast cancer, CMF consisted of cyclophosphamide (100 mg/m2
orally from day 1 to 14), methotrexate (40 mg/m2
intravenously on days 1 and 8), and fluorouracil (600 mg/m2
intravenously on days 1 and 8), repeated every four weeks for either six or 12 cycles.2
In this subset of patients, women older than 60 were to receive reduced doses of methotrexate (30 mg/m2
) and fluorouracil (400 mg/m2
In the third study, 12 cycles of cyclophosphamide (600 mg/m2
), methotrexate (40 mg/m2
), and fluorouracil (600 mg/m2
) were given intravenously on day 1 and repeated three weeks later. No dose reductions for older patients were planned.9
In all studies, treatment with CMF was started two to four weeks after surgery. No other adjuvant treatments, in particular no endocrine treatment, were allowed, with the exception of breast irradiation for patients who had had conservative surgery. Breast irradiation (40 Gray (Gy) plus a boost of 10 Gy in four to six weeks) had to be initiated within six to eight weeks from surgery and was administered alongside CMF in women allocated to receive adjuvant chemotherapy.
Details on baseline studies and follow up programmes are reported elsewhere.2,7-9
Before surgery, all patients had a complete physical examination and were investigated by radiological and biochemical techniques. In the absence of symptoms, women had a complete physical examination every three months during the first year, every six months for the next four years, and every 12 months for the following 10 years. Biochemical tests and radiological studies were done every six to eight months during the first five years, and once a year thereafter. Liver ultrasonography was undertaken only if clinical or biochemical findings were suspicious. Mammography, bilateral in patients who had had conservative surgery, was scheduled once a year. After the 15th year of follow up, examinations were set for every 12 to 24 months, and when these were not done in the outpatient clinic of the institute, contact was periodically maintained with the patients and their family doctors. Patients with suspicious or controversial findings were examined more often. We considered treatment to have failed when the first evidence of new manifestations of disease in locoregional areas (including ipsilateral supraclavicular adenopathy), distant sites, the contralateral breast, or any combination of these sites was documented. We considered neither second primary cancers nor deaths owing to causes other than breast cancer treatment failures.
We calculated relapse free survival from the date of surgery to the first documented evidence of treatment failure. We used death from all causes as the end point for overall survival, which we also measured from the date of surgery. We analysed whether drug induced amenorrhoea in women menstruating at study entry was able to influence the outcome of treatment, excluding all patients who had a relapse within the first nine months after surgery.10
We used the Kaplan-Meier product limit method for all survival analyses.11
We used the log rank test to test the null hypothesis concerning the differential effects of treatment or of some prognostic factors in univariate analyses,12
and all P values were two tailed. In addition, we used a Cox regression model to investigate the joint effects of treatment and of prognostic indicators,13
using a backward selection procedure. We used Wald statistics to test the null hypothesis of the regression analysis.14
We estimated the relative risks as hazard ratios and calculated the rate of the sites of disease relapse as first event according to the method proposed by Marubini and Valsecchi.15
We analysed the data that were available as of 28 February 2003. Only two patients in complete clinical remission were lost to follow up, one after 15.8 years and the other after 20.7 years.