K.D. Macrae

A comparison of doxorubicin, epirubicin and mitozantrone as single agents in advanced breast carcinoma

Between July 1985 and December 1987, 87 patients with advanced breast carcinoma were randomized to receive single agent doxorubicin (70 mg/m2), epirubicin (70 mg/m2) or mitozantrone (14 mg/m2) at 3-weekly intervals. The patients had received no previous chemotherapy for their advanced disease but 91% had received prior hormonal therapy. The response rates were 36% with doxorubicin, 32% with epirubicin and 26% with mitozantrone, but these differences did not reach statistical significance. The median survival of all patients was 8.3 months. There was no significant difference in response rates or survival according to menopausal status. The toxicities of the three agents are compared. Nausea, vomiting and alopecia were more severe in patients treated with doxorubicin or epirubicin than those treated with mitozantrone. Myelosuppression and infective episodes occurred more frequently with mitozantrone. Two cardiac complications were reported. This study shows that the toxicity and low efficacy of all three agents limit their use as single agents in advanced breast carcinoma. The role of single agent chemotherapy and the relative toxicities of these drugs are discussed.

COMPARISON OF 5-DAY, 1-DAY, AND 2-DAY CYCLICAL COMBINATION CHEMOTHERAPY IN ADVANCED BREAST CANCER

The results of 5-day cyclical combined chemotherapy for advanced breast cancer were compared in two trials with the less demanding 1-day and 2-day regimens. At all stages, except at 3 months, 5-day treatment was significantly more successful in providing regression and remission of tumours than the 1-day regimen. However, although the differences are not statistically significant, data from the second trial indicated that the 2-day regimen tended to be more effective than the 5-day regimen. The lower 95% confidence limit for the success of the 2-day treatment was 56.2% at 3 months.

A randomized, prospective, comparative, multicentre trial of a single combination versus alternating combinations of antitumour drugs in advanced breast cancer

A prospective, multicentre trial was conducted in 262 patients with advanced breast cancer randomized to receive every 3 weeks either: (i) a single five-drug combination of adriamycin (50 mg), vincristine (1.5 mg) and 5-fluorouracil (750 mg) given intravenously; with methotrexate (50 mg) intramuscularly and chlorambucil (10 mg) orally all at time zero, followed by three further doses of chlorambucil (10 mg) at 6-h intervals, or (ii) one course of two alternating three-drug combinations consisting of regimen A--vincristine (1.5 mg), adriamycin (70 mg) and methotrexate (50 mg) and regimen B--5-fluorouracil (750 mg) and vindesine (5 mg) intravenously with cyclophosphamide (50 mg) orally at time zero, followed by three further doses of cyclophosphamide (50 mg) at 6-h intervals. Results show that overall response rates to chemotherapy were comparable in the two arms of this study being 63% (83 of 131 patients) for the single combination and 64% (84 of 131 patients) for the alternating combinations. Response rates according to menopausal status indicate no significant difference for the two arms of this study. However overall, combining all patients treated with either the single or the alternating combinations, post-menopausal patients had a significantly lower response rate (57%) compared with pre-menopausal patients (76%), P less than 0.05. Overall serious side-effects were minimal and were similar in both treatment groups. Response durations and overall survival data, which are essentially similar for the two treatment groups, proved disappointing with a median response duration of only approx. 6 months and overall median survival only slightly in excess of 1 year. Alternative treatment approaches are needed to maintain the remissions initially achieved in metastatic breast cancer.

Treatment of Early Breast Cancer

SIR,-From the evidence in the excellent paper by Sir Hedley Atkins and others (20 May, p. 423) it would appear that there is no case for the operation of extended tylectomy in the management of stage I and II carcinoma of the breast in patients over 50 years of age. Before all forms of local surgery in carcinoma of the breast are abandoned several questions might be asked conceming extended tylectomy. In some patients a more adequate margin that the stated 3 cm can be easily removed without distortion of the breast contour. Was more than this margin ever excised? If skin dimpling is present is skin removed? In the subareolar lump is the nipple removed? Answers to these questions would help in the evaluation of extended tylectomy. Spread of carcinoma within the breast is by periductal lymphatics and therefore to eradicate the disease from within the breast excision of breast tissue on a lobular basis is logical. Support for this concept is obtained by whole organ sectioning, a laborious task. In breasts removed for primary carcinoma intraduct changes of epitheliosis, in-situ carcinoma, and frank invasion are commonly seen and are most marked in the lobule containing the carcinoma. When there is clinical evidence of skin involvement an adequate clearance can be obtained only by excising skin with the tumour. Surely there are many factors which influence the surgeon in his assessment of how much tissue to excise around the tumour. To stick rigidly to a fixed margin clear of the tumour is not always as appropriate as adopting a more flexible policy (according to various clinical criteria). It is of paramount importance to excise an adequate margin, which should be at least 3 cm from the edge of the tumour, as measured from the macroscopic appearance of the bisected lump. This policy is more reliable than taking the measurement from the palpable tumour edge. If skin tethering is present a good margin of skin clear of the tethered area should be included in the .excision. In the well-endowed breast it is easy to excise more tissue and this should be done, in each case removing the segment of the breast affected in order to remove the involved lobule in its entirety. Not often discussed is the relative size of the lump to the breast. This ratio is often a more pertinent factor in the formation of a treatment policy, particularly with regard to the small lump in a large breast, than the overall palpable tumour size such as is used in Manchester staging. Great variation in breast size exists between individuals as between nations such as Japan and Italy. An adequate clearance in the Tokyo breast is not as easily achieved as a similarly adequate clearance in its Neapolitan counterpart. Incidentally the word tylectomy combines both Greek and Latin roots and therefore is no improvement upon lum,pectomy.-We are, etc., J. W. S. RICKETT A. G. NASH Royal Marsden Hospital, London S.W.3

‘Early’ and ‘late’ breast cancer: A unified concept for treatment

Most women with early breast cancer have distant metastases by the time the primary growth comes to diagnosis. This observation is based upon the fact that these frequently appear despite successful removal of the primary growth, and given that they originate from the now no longer present tumour, the inescapable conclusion is that dissemination must have taken place prior to initial treatment. Failure to appreciate this rather obvious point results from shortcomings in available diagnostic technology, and inadequate usage of that which is available. Whether detected early or late, treatment of this disseminated cancer poses a common problem, and the various systemic methods are reviewed. Androgens, oestrogens, progestogens and single drug chemotherapy can, from time to time, produce useful results. By careful selection of patients with appropriate metastatic patterns, these methods may yield improved response rates up to 50% but, by and large, experience remains disappointing with across the board representative figures of perhaps 25% response being commonplace. Additionally, these methods are not without their side effects which can be distressing, and indeed on occasion life-threatening. Their short-comings have led to the development of cyclical combination chemotherapy as here reported. Cyclophosphamide, vincristine, 5-fluorouracil and methotrexate are administered for five consecutive days per month. The results obtained in 100 patients treated with this technique are compared with the authors previous experience with norethisterone acetate and hypophysectomy. Whether the results are considered in total or broken down into prognostic categories such as predominant metastatic pattern, disease-free interval, or age, combination chemotherapy has a clear advantage over other techniques. Whilst this experience does not constitute a randomised trial, it is considered that the results are so superior to previous treatment techniques as to render such a study unnecessary or even unethical. It is proposed that combination chemotherapy should constitute the first line of treatment for overt advanced breast cancer. It is further proposed that its role should be explored at an earlier stage in the treatment of breast cancer. Before this can take place however, treatment must be made less demanding and less toxic. Further research is underway with these aims in mind.

Some Remarks on Edwards' Probability-Preference Experiments

A series of experiments by Edwards (1953, 1954a, 1954b) was presented as demonstrating the existence of probability-preferences in gambling and has been quoted in review articles (Edwards, 19%; Luce & Suppes, 1965) as having shown the importance of these probability-preferences. Edwards (1953) constructed three sets of bets, one set having the same positive expected value (PEV), one set having zero expected value (ZEV), and one set the same negative expected value (NEV). The bets were twooutcome bets, and the expected value (EV) of a rwo-outcome bet is found by EV = PI VI + Pa V?, where P1 is the probability that outcome 1 occurs, P, is the probability that of outcome 2. Edwards PEV bets had an EV of 52.5 cents, e.g., he used a bet having a probability of 0.25 of winning

SERIAL POSITION AND SEQUENTIAL DEPENDENCIES IN REPEATED MEASURES DESIGNS

2.10 and a probability of 0.75 of winning nothing. All his PEV bets were of this form, that is, having one positive outcome and one zero outcome. His NEV bets were obtained by substituting the word lose for the word win in these PEV bets. Thus, the NEV bet obtained from the above PEV bet would be a probability of 0.25 of losing

TREATMENT OF OVARIAN CARCINOMA

2.10 and a probability of 0.75 of losing nothing. Thus, all his NEV bets were of the same form, having one negative outcome and one zero outcome. Edwards ZEV bets had, of course, to involve both positive and negative outcomes. For example, he used a bet having a 0.25 probability of winning

MULTIMODAL THERAPY FOR STAGE-II BREAST CANCER

2.10 and a probability of 0.75 of losing

Separation of Normals, Neurotics and Simulating Malingerers on the MPI by means of Discriminant Function Analysis

0.70. These three sets of bets are of very different sorts. Only the ZEV bets are of the type usually found in gambling situations, where it is possible to win or to lose on any bet. Edwards PEV and NEV bet probability-preferences cannot therefore be generalized to the more usual win-or-lose sort of bet found in gambling. Perhaps a more serious criticism of the Edwards studies is that his interpretation of preferences among his bets as being probability-preferences is questionable. Edwards said that his PEN bets showed preferences d o n g probabilities of winning, that his ZEV bet preferences also showed preferences among probabilities of winning, and that his NEV bet preferences showed preferences among probabilities of losing. This interpretation ignores the fact that his ZEV bets contain, as they indeed must, both a probability of winning and a probability of losing. Which probability is S preferring when he chooses a ZEV bet? Further, preferences among the PEV bets could be preferences for probabilities of not winning, and preferences among the NEV bets could be preferences for probabilities of not losing. Also, Edwards does not argue why his bet preferences indicate probability preferences and not subjective probabilities or a departure from linearity in the utility function on the outcomes involved. The nature of the bets and the interpretation of the results make these studies inadequate.