Article Text

Download PDFPDF
97 Overdiagnosis and overtreatment of colorectal lung metastases: a study of the (lack of) impact of controlled trial evidence
  1. Norman Williams1,
  2. Francesca Fiorentino2,
  3. Fergus Macbeth3,
  4. Tom Treasure1
  1. 1UCL, London, UK
  2. 2KCL, London, UK
  3. 3Cardiff University, Cardiff, UK


Objectives Our objective was to evaluate published responses to the Pulmonary Metastasectomy in Colorectal Cancer (PulMiCC) randomised controlled trial (RCT); preliminary results of this work were presented to Preventing Overdiagnosis in Sydney in 2019. Clinical selection for operation was made on the basis of prognostic factors: number of metastases, interval since primary resection, tumour markers, liver involvement, age, performance status and lung function. When well-balanced in an RCT, median (95% CI) survival after metastasectomy was 3.5 (3.1– 6.6) years compared with 3.8 (3.1–4.6) years for controls. ‘Expert Consensus’ from the Society of Thoracic surgeons (2019) assumed five-year survival without metastasectomy would be nil thus allowing all metastasectomy survival to be attributed to surgery. In fact, survival was 30% (15%–46%) for RCT controls and 22% (15%-30%) among 128 non-randomised patients selected to not have metastasectomy. These results confront the widely believed large survival benefit from local treatment of lung metastases.

Method Published articles that cited the PulMiCC trial were identified from Clarivate Web of Science©. Duplicates and self-citations were excluded, and relevant text extracted without authors’ names or affiliations. Four independent researchers with evidence-based medicine training and credentials, but no involvement with the PulMiCC trial, agreed to rate the published responses using mutually agreed numerical ordinal 0–6 scales for the representativeness of trial data and the textual tone. Ratings for data ranged from no data to a full and fair summary. For tone they were from dismissive to supportive with a tipping point of 3 for a balanced appraisal. After a first independent round the ratings were tabulated with an analysis of dispersal; the spreadsheet was returned for a second round where the researchers could reconsider ratings. The ratings were aggregated and summarised.

Results Sixty-four PulMiCC-citing publications were identified, and relevant text was extracted. The consensus rating for data inclusion was a median of 0.25/6 (range 0–5.25, IQR 0–1.5) and for textual tone the median rating was 1.87/6 (range 0–5.75, IQR 1–3.5). The majority (57/64) of citations did not provide adequate representation of the PulMiCC data and 31 gave no data. The overall textual tone was dismissive. Many discounted the findings because the trial closed early and was ‘underpowered’. Due to the size of the RCT it was unable to show non-inferiority of non-operative treatment, but authors appeared to use this as a reason to discard all the data as of no value. Two misinterpreted the authors’ conclusions as supporting metastasectomy. Those who dealt with the findings at any length noted and accepted that five-year survival without operation was much higher than widely assumed.

Conclusions It was already known from two meta-analyses of 16 RCTs that earlier detection of metastases by intensification of monitoring did not improve survival. Direct evidence from PulMiCC showed no survival benefit but documented harm and deaths after surgery. Due to resistance to randomisation PulMiCC was underpowered to test non-inferiority, but the results in 391 elective patients, and 93 in the nested RCT provide trustworthy evidence to undermine the widespread belief in a large survival benefit from metastasectomy. The present study of published reactions reveals a widespread reluctance to accept the findings of a carefully conducted RCT which is likely due to ‘motivated reasoning’, and a refusal to accept the evidence that this medical intervention represents overdiagnosis and over treatment. Journal reviewers and editors should be wary of accepting manuscripts that selectively discuss the results of research in a biased manner.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.