In contrast to the well established prognostic factors determining outcome in invasive breast carcinoma [11
], the value of similar prognostic indices has proved less clear cut in DCIS. The present study of 215 patients with DCIS treated with local excision and observation alone is one of the largest series in which recurrence is unaffected by radiation therapy, hormone manipulation or chemotherapy and has given us the opportunity to assess the prognostic impact of patient and tumour characteristics free of any potentially confounding treatment related influences.
In this study we have shown that for those patients with a low VNPI score (scores 3–4, n = 61) the recurrence rate and hence the chance of developing invasive breast cancer is minimal (0% over 8 years, P = 0.002). These patients we feel should not receive RT. For those with intermediate (scores 5–7, n = 104) and high (scores 8–9, n = 20) VNPI scores the chance of developing any recurrence over 8 years in this study is 21.5% and 32.1% respectively (P = 0.002). Taking these factors in to account and appreciating that the natural history of DCIS remains elusive, it is our opinion that RT should be reserved for those patients with high and possibly intermediate VNPI scores as it is in these groups that the benefit: risk ratio is likely to be highest.
The effect of including the small number of patients with tumours that did not have their margin width recorded (n = 18) in the analysis of the VNPI's effect on disease free survival would re-enforce its significance, as all had low scores (3–4) and in none of the cases was a recurrence observed.
Comedo necrosis was found to be present in 84 cases (39%) and when analysed in combination with grade of tumour, as specified in the VNPI, was found by univariate analysis to adversely influence disease free survival (p < 0.05). In Cox multivariate regression analysis, none of the individual components of the VNPI reached statistical significance, suggesting that the whole Index is of greater value than its parts. Adding age to the index reduced rather than increased its prognostic value.
Obviously it is important to note that the retrospective nature of this study means that conclusions must be drawn with caution. There is currently a wealth of relatively small series of studies and personal opinions regarding the decision to give or withhold RT as a primary treatment measure in DCIS [2
]. Results and opinions are often conflicting. Advocates for giving this modality point to the fact that the only level I evidence that is available, the gold standard in today's evidence-based practice, demonstrates without question that RT reduces local recurrence [4
]. Furthermore it has been suggested that the reason why a survival benefit has not been demonstrated in the large randomised trials is due simply to the fact that the follow up period has not yet been long enough [3
In contrast there are clinicians on both sides of the Atlantic who feel the methodology of the aforementioned trials, especially regarding the measurement of margin width which has been shown by certain authors to be a determinant of local recurrence [14
], raises concerns about the significance and therefore applicability of the results. Those who are reluctant to use RT for DCIS as a primary treatment argue that a substantial proportion of lesions behave in a benign fashion and are unlikely to transform into carcinoma during the patient's life-time [15
] and as such it is unreasonable to indiscriminately subject the increasingly large number of women with screen detected DCIS to the potentially serious side effects of RT, when such therapy has yet to demonstrate a survival benefit.
Perhaps the most convincing evidence against adopting such a stance has been described by Wong et al
]. These authors conducted a single arm prospective trial evaluating recurrence rates after breast conserving surgery alone in a group of patients in which they predicted that the rate of recurrence would be low (margins >1 cm, low/intermediate grade DCIS). The trial was prematurely stopped after the predefined boundaries for what was deemed as an acceptable recurrence rate was overstepped. The estimated 5 year ipsilateral local recurrence rate in the 158 patients accrued was 12%, which is a value similar to the surgery only arms of the UKCCCR, EORTC and NSABP trials [4
] and as such appeared to support the conclusion that there is in fact not a subgroup of patients with DCIS, for whom RT should not be offered.
Silverstein and Lagios [2
] have highlighted various factors in the methodology of this study which may partially be responsible for the relatively high recurrence rates observed. They also point out that the majority of cases of recurrence were non invasive (69%) in nature and could be treated by re-excision plus or minus RT with an expected 100% cause specific survival. They further calculate that taking into account the cases of invasive recurrence (31%) the expected cause specific mortality at 12 years would be only 0.6% and consequently the harm avoided by withholding RT in 158 patients should result in this trial being viewed not as a failure but rather as a success.
More recently Macausland et al
] made an attempt to validate the VNPI but found that although trends were observed between this stratification system and local recurrence, none reached statistical significance. A significant number of patients in this cohort received tamoxifen as adjuvant therapy however and this may have influenced results. Additionally the authors acknowledge that the predictive utility of the VNPI in this study may well be seen with further follow-up.
As a consequence of the controversy surrounding the decision whether to give or withhold RT, there is a substantial lack of standardization in the treatment for DCIS at both national and international level [17
]. It seems that until there is sufficient level I evidence determining that a certain subgroup of patients who, following wide local excision alone, are shown to have a rate of recurrence that is less than or at least equal to those described in the surgery plus RT arms of the large trials a lack of uniformity will persist. Whether identification of such a subgroup, if it does indeed exist, is to be made using a relatively simple scoring system such as the VNPI, or by the detection of more advanced biological markers is not yet clear [18