Dear Editor, We are writing to bring serious concerns about the validity of the article "Hematologic toxicity assessment in solid tumor patients treated with cetuximab: A pooled analysis of 18 randomized controlled trials" 1 authored by Ran Cui et al. and recently published by the International Journal of Cancer to your attention. After careful reading, we firmly believe that the review uses inappropriate methods and incorrect data, and therefore that the conclusions are misleading.We should be clear that the reason that we have been able to consider the data so carefully is that our group is conducting a systematic review on serious thromboembolic adverse events which has a very similar design (PROSPERO registration number CRD42014009165 2 ), although our systematic review also includes randomized controlled trials (RCTs), in which panitumumab (another approved anti-EGFR monoclonal antibody) is tested as an add-on to standard regimens.We detail the many shortcomings of the research reported by Cui et al. below.
Search Strategy and Missing TrialsWe were surprised by the small number of records (511) returned by the authors' initial search, the number of RCTS identified (18) and by the small number of articles that were excluded on grounds of lack of data (2). In contrast, our own search yielded a total of 6777 records of which 45 3-5,7-48 were RCTs. Of the 37 [3][4][5][6][7][8][9][10][11][12][13][14][15][16]19,20,[22][23][24][25][26][27][28][30][31][32][33][34][36][37][38]40,41,43,45,47,48 trials of cetuximab, 32 [3][4][5][6][8][9][10][11][12][14][15][16]19,20,[22][23][24][25][26][27][28][30][31][32][33][34][36][37][38]40,41,43,45,47,48 reported data on hematologic toxicities.Cui et al. do not report their search strategies in full, making it difficult to have confidence in their quality. From the information given, it would seem that although there are several rigorous search filters that can be used to identify RCTs, none of these were used in either MEDLINE or Embase. It also seems that the Cochrane Central Register of Controlled Trials was not used. The impact of these choices would be to restrict the number of articles identified by the literature searching.As of October 31, 2013, the time at which the literature search by Cui et al. was done, 6 of the additional 16 RCTs that we identified were already published. These 6 trials (including over 1,700 participants) should therefore have been identified and added to the 18 trials that they included (Table 1).To date, further data from >5,000 patients are available, as compared to the analysis of Cui et al.; around one-third of these data were already published at the time of their search, and therefore have been missed from the review.