Prognostic Relevance of Palliative Primary Tumor Removal: Why Randomized Controlled Data Are Still Mandatory
With great interest, we read the results of the population-based study on the prognostic relevance of primary tumor removal by Tarantino et al.1 The authors show a clear association between resection of the primary tumor and improved overall survival that is sustained in a Cox regression analysis after propensity score matching. These data are in line with our data from 2 large phase III trials, which indicate that resection of the primary tumor is a prognostic factor for survival in stage IV colorectal cancer.2 Although this is one of the largest studies on this subject using data from randomized controlled trials, the authors surprisingly did not include it in their reference list. Although we agree with the authors that it is unlikely that this large difference in overall survival can be solely contributed to confounding, we would like to respectfully raise some issues that we believe to be insufficiently highlighted by the authors.
First, the SEER database does not provide information on systemic treatment. Likewise, no data on performance status and/or comorbidity are available, making it extremely difficult to estimate the number of patients within the study population that received systemic therapy. That this is extremely important, because it shows the difference in overall survival between patients who do and do not receive systemic therapy. Patients who do not receive systemic therapy have an estimate overall survival of about 8 months,3 whereas recent trials report median overall survival of more than 30 months for patients treated with systemic therapy.4 Patients who are eligible for palliative resection of the primary tumor will often also be eligible for systemic therapy, as many of the criteria (eg, good physical status, absence of severe cardiovascular disease, and younger age) are similar for both treatments. This leads us to believe that the use of systemic therapy might have been unequally distributed between the 2 groups. Indeed, patients’ age, the only provided characteristic that is clearly associated with systemic treatment,5 is significantly different between the 2 groups. The probability of this selection bias is also supported by the reported median overall survival of the nonoperative group (3 to 7 months), which is considerably lower than reported for nonoperative management in other studies.2 This potential difference in systemic therapy might account to some extent for the difference in overall survival. More accurate results could have been obtained if patients who only received best supportive care were excluded from the nonoperative group.
Second, we would like to emphasize that other outcomes should be considered in the discussion about palliative resection of the primary tumor. Quality of life, an important issue in this discussion, is completely neglected by the authors. In our opinion, a survival benefit of at least 6 months is necessary to balance out surgery-related impact on quality of life. However, it should be taken into account that individual patients might have a different opinion depending on their personal values. Furthermore, as the authors briefly mention in the introduction, an important argument against surgical management is the associated postoperative complications that may lead to a delay or even cancellation of systemic therapy. Although the SEER database does not provide information on postoperative complications and morbidity, we wonder why the authors did not report 30-day mortality rates to give a pointer on this specific topic.
Last, the authors specifically address the recommendation by the National Comprehensive Cancer Network that surgery should only be performed in asymptomatic patients with synchronous colorectal metastases if performed in a curative attempt. However, no efforts are done to estimate the proportion of asymptomatic patients in the study population.