Thrombocytosis: an important marker of cancer in primary care
Thesis or dissertation
University of Exeter
Reason for embargo
Papers currently in press in journals that require embargo
Thrombocytosis (raised platelet count) has recently been identified as a prediagnostic risk marker of cancer; however, the association has not been fully investigated. This thesis aimed to explore the relationship between thrombocytosis and a future diagnosis of cancer through three complementary pieces of research. Firstly, a systematic review was carried out which aimed to identify studies that had investigated thrombocytosis as a diagnostic marker of cancer. Four case-control studies were identified that had found thrombocytosis to be a significant predictor of lung, oesophago-gastric, uterine, and renal cancer. A further four studies found that thrombocytosis did not predict pancreatic, breast, ovarian, or colorectal cancer. One further study had collected, but not analysed, platelet count data. Data from all nine studies were included in a meta-analysis. The findings of the review suggest that thrombocytosis is a marker of some, but not all, types of cancer. The second study used data from the Clinical Practice Research Datalink (CPRD) and the English cancer registry. This cohort study examined the relationship between thrombocytosis and cancer using two groups of patients. The first included 40,000 patients with a raised platelet count (a platelet count of > 400 x 109/L). The second cohort included 10,000 patients with a normal platelet count (150 - 400 x 109/L) who were age, sex, and practice matched to a random quarter of the first cohort. This study found that the risk of cancer was greater in patients with thrombocytosis compared to those with a normal platelet count. The one year cancer incidence was 11.6% (95% CI 11.0 - 12.3) for male patients with thrombocytosis, and 4.1% (95% CI 3.4 - 4.9) in males with a normal platelet count. In female patients, the one year cancer incidence was 6.2% (95% CI 5.9 - 6.5) for those with thrombocytosis and 2.2% (95% CI 1.8 - 2.6) for those with a normal platelet count. Lung and colorectal cancer were more likely to be diagnosed in patients with thrombocytosis than in patients with a normal platelet count, and breast and prostate cancer less likely. In patients with a sustained increase in platelet count over six months, the risk of cancer increased to 18.1% in males (95% CI 15.9 - 20.5) and 10.1% in females (95% CI 9.0 - 11.3). Around a third of patients with lung or colorectal cancer and thrombocytosis had no other symptoms prior to diagnosis that would have prompted investigation for cancer as per current NICE guidance. The third study compared cancer recording in the CPRD and in the English cancer registry. The aim of this study was to examine the validity of cancer recording in the CPRD using cancer registry recording as the gold standard, and to estimate predictors of concordance between the two data sources. A sensitivity analysis repeated the primary analysis from the second study to estimate the effect of including unverified CPRD cancer diagnoses. The CPRD identified 5,924 of 7,785 cancers recorded in the cancer registry (sensitivity 76.1%, 95% CI 75.1 - 77.0). 36,255 patients with no record of cancer in the CPRD also had no cancer record in the cancer registry (specificity 97.0%, 95% CI 96.1 - 97.2). 5,924 of 7,028 CPRD cancer diagnoses were confirmed by the cancer registry data; the positive predictive value (PPV) of a CPRD recorded diagnosis was 84.3% (95% CI 83.4 - 85.1). Male cancers, those in younger patients, and those recorded from 2005 onwards were more likely to be recorded in both sources. In a sensitivity analysis, the exclusion of cancer diagnoses that were only recorded in the CPRD did not significantly alter findings from the cohort study described above. The findings from this thesis show that thrombocytosis is an important predictor of undiagnosed cancer in adults aged 40 years and over. Patients with thrombocytosis are more likely to be diagnosed with lung and colorectal cancer than other types. These results suggest that cancer should be considered as an underlying diagnosis in patients with unexpectedly raised platelets, even if cancer was not suspected at the time that the blood test was ordered. For at least a third of patients with thrombocytosis and cancer, there will be no other clinical features of malignancy; for this proportion, thrombocytosis has great potential to expedite diagnosis and improved survival.
PhD in Medical Studies