Note from the National Guideline Clearinghouse (NGC): The National Institute for Health and Clinical Excellence (NICE) commissioned an independent academic centre to perform a systematic literature review on the technology considered in this appraisal and prepare an assessment report. The assessment report for this technology appraisal was prepared by the West Midlands Health Technology Assessment Collaboration (see the "Companion Documents" field).
Search Strategy
The search strategy was divided into 6 parts and aimed to look for trials of imatinib (with or without standard treatment comparators), trials of alternative/experimental treatments, studies that had observed patient prognosis without treatment (to enable a comparison of disease progression should trials without comparators be available), and diagnostic papers in order to gain an insight into the uncertainty of gastro-intestinal stromal tumour (GIST) diagnosis and possible consequences of treating false positives. In addition ongoing trials were sought, as imatinib is a very recent drug. A search for economic evaluation of treatments for GIST was also conducted.
The searches were not restricted by language. Published and unpublished studies were sought. Databases were searched from inception. Searches (except for ongoing trials) were undertaken between 25 April and 15 May 2003.
See the section 2 of the assessment report for details of the electronic search.
Inclusion and Exclusion Criteria
A three stage sorting process was instigated to look through the yield of the search.
Stage 1 - Including or Excluding Studies
Two reviewers independently assessed papers for inclusion/exclusion using the title and where available the abstract. The following inclusion criteria were applied:
Inclusion Criteria
Study design: Relevant randomised controlled trials (RCTs), non-randomised controlled studies, cohort studies, and case series that reported effectiveness results of treatment with imatinib and/or other interventions in patients with GIST.
Population: Ideally patients diagnosed cKIT positive unresectable and/or metastatic GISTs (including primary or recurrent tumours). Not so ideal but still included were patients histologically diagnosed with GIST. In trials older than 1999 patients who were diagnosed with gastrointestinal leiomyosarcoma or soft tissue sarcoma that appeared to behave as GIST (e.g., tendency to metastasize in the liver), were included. Early terms for GIST4 could include oesophageal leiomyosarcoma; gastric leiomyoma; gastric leiomyoblastoma; small intestinal leiomyoma and leiomyosarcoma; colonic and rectal leiomyoma and leiomyosarcoma; gastrointestinal autonomic nerve tumour (GANT); leiomyoma and leiomyosarcoma of omentum and mesentery; retroperitoneal leiomyosarcoma.
Intervention: Imatinib. Oral dosage -- any dose. (Where imatinib = STI 571, Glivec, Gleevec, or CGP57148).
Comparators: The ideal comparator was the current standard treatment (symptom-relief and best supportive care), or placebo. If there were no trials with these comparators, data from trials that investigated experimental treatments in patients with GIST were sought, so that an indirect comparison could be made.
Outcomes: See the "Major Outcomes Considered" field in this summary.
Disagreements were resolved by discussion. Inclusion/exclusion decisions were made prior to detailed scrutiny of the results and study quality assessment. Foreign language publications were screened using English abstracts where available.
Stage 2 Consensus Meeting
Because the initial systematic search and sort at stage 1 had yielded in excess of 1000 papers using the above criteria, it was felt that tighter criteria were needed to eliminate papers that could not add substantial value to the review. In particular a large yield had come from prognosis/natural history papers and diagnostic papers. It was therefore agreed that the following inclusion criteria were to be applied:
Imatinib effectiveness - any patient with GIST (at any stage) who has been treated with imatinib. Ignore reviews and case studies of single patients published in abstract form only.
Other treatments -- any patient with GIST (at any stage) who has been treated with drugs other than imatinib, also include other procedures (e.g., surgery, radiotherapy, brachytherapy). Exclude papers that compare surgical laparoscopy vs. open surgery.
Prognosis -- papers describing primary research that involved the prognosis of 10 or more patients where clinical outcomes are described. Ignore reviews.
Diagnosis -- papers describing primary research that involved 10 or more patients with clinical outcomes reported. Major reviews on diagnostic accuracy or diagnostic criteria of GIST, especially those describing advanced disease were included.
Three reviewers applied the criteria on the papers selected at stage 1, and disagreements were resolved by discussion.
Stage 3
Full paper copies of studies identified in stage 2 were obtained for detailed examination. At this stage, additional papers were excluded as and when detailed study of the methods revealed that the paper did not meet the inclusion criteria. Usually this was because the wrong populations had been used; in particular some papers on examination had used patients with primary disease that was treatable with surgery and was not metastatic. Translations were also obtained on full papers where necessary or where possible. Translations were not obtained for 4 case studies included in the review, as it was not felt that a translation would add value to the review.
Data Extraction Strategy
Two reviewers independently extracted data using a pre-designed data extraction form (see Appendix 2 page 81 in the assessment report). Disagreements were resolved by discussion, consulting with a third party where necessary. Where there was missing information and time constraints allowed, the authors were contacted. Data from studies with multiple publications were reported as a single study but the source of the publications was noted.
Quality Assessment Strategy
Quality of studies was assessed using the York CRD criteria for experimental and observational studies (Appendix 11, page 128 in the assessment report). These criteria were tested and revised where necessary. The following quality issues were felt to be of paramount importance: study design, patient characteristics, (in terms of GIST diagnosis, disease severity, length of time with GIST), and any possible sources of biases in patient selection, treatment provided, and outcomes measured; where found these were reported.