With interest, we read the article by Hwang et al about survival after lumpectomy and mastectomy for early stage invasive breast cancer. The authors reported better overall survival and disease-specific survival in patients who underwent lumpectomy and received radiation (breast-conserving therapy [BCT]) compared with patients who underwent mastectomy. However, we wish to make a few remarks regarding the interpretation of these results.
By comparing 2 treatment modalities in an observational study, the authors induced the bias of “confounding by indication.” In observational data, it cannot be recalled why a patient receives a specific treatment. For example, the authors demonstrate that the patients who underwent mastectomy generally were older, had higher tumor stage, more often had grade 2 or 3/4 disease, more often had hormone receptor-negative tumors, had more positive lymph nodes, and had larger tumors. All of these characteristics are obviously associated with poorer survival rates. Although the authors adjusted for these registered differences in patient and tumor characteristics, there are still unregistered residual confounders that can be of influence in the choice of a certain treatment.
The authors address this problem in the discussion, and argue that they have performed several analyses to counter this issue. However, by demonstrating that patients who underwent mastectomy probably had a greater comorbidity burden, they even endorse the assumption that this study suffers strongly from confounding by indication. This also explains the largest survival benefit in patients aged 50 years and older with hormone receptor-positive disease who received BCT, because elderly patients who have multiple comorbidities and poor physical functioning will undergo mastectomy more often and will have worse overall survival.
Observational studies are especially important in the elderly breast cancer population, because the elderly are under-represented in clinical trials due to restrictions based on age, comorbidity, and functional status. However, clever study designs are needed if one wants to draw conclusions about treatment effects. It is not possible to compare treatment modalities in a retrospective observational study unless it is possible to use an instrumental variable, in which a factor is available that determines treatment allocation but is unrelated to the outcome. As far as the evidence reaches, breast cancer outcomes are equivalent with either mastectomy or BCT, and patients with early stage breast cancer should be given the opportunity to make a well balanced choice for either 1 of these treatment modalities.
Nienke A. de Glas, MD1
Mandy Kiderlen, MD1,2
Gerrit-Jan Liefers, MD, PhD1
1Department of Surgery
2Department of Gerontology and Geriatrics, Leiden University Medical Center, Leiden, the Netherlands