Blood CEA levels for detecting recurrent colorectal cancer: a Cochrane Diagnostic Test Accuracy review

Conference: 
SAPC ASM 2015 - Oxford - Evidence and innovation in primary care
Talk Code: 
1A.2

The problem

Blood carcino-embryonic antigen (CEA) testing is a recommended part of follow-up to detect recurrence of colorectal cancer following primary curative treatment. As a simple blood test, CEA could be used in primary care reducing the cost of follow-up and the risk to the patient. Whilst 5µg/L is most commonly reccommended there is substantial clinical variation in the cut-off level applied to trigger further investigation. We aimed to determine the diagnostic performance of different blood CEA levels in identifying patients with colorectal cancer recurrence in order to inform clinical practice.

The approach

We followed Cochrane Diagnostic Test Accuracy Review methodology. Our searches retrieved 6827 citations. 268 full-text articles remained following title and abstract screening by three authors. 52 studies reporting the diagnostic accuracy of CEA for the detection of colorectal cancer recurrence and reporting a 2x2 table compared to an appropriate reference standard were included in the final review. All included studies were quality assessed using a modified QUADAS-2 tool. Forest and ROC plots were used to summarise the findings.

Findings

In the 52 included studies, sensitivity ranged from 41% to 97% and specificity from 52% to 100%. In the 7 studies reporting the impact of applying a threshold of 2.5µg/L, pooled sensitivity was 82% (95% CI: 78-86%) and pooled specificity 80% (95% CI: 59-92%). In the 23 studies reporting the impact of applying a threshold of 5µg/L, pooled sensitivity was 71% (95% CI: 64-76%) and pooled specificity 88% (95% CI: 84-92%). In the 8 studies reporting the impact of applying a threshold of 10µg/L, pooled sensitivity was 63% (95% CI: 53-73%) and pooled specificity 97% (95% CI: 90-99%). Assuming a 2% prevalence of recurrence in each 3-6 month time-period between tests, the false alarm rate would be approximately 9 out of 10 (92% and 89%) applying thresholds of 2.5µg/L and 5µg/L but could be reduced to 2 out of 3 (69%) by applying a threshold of 10µg/L.

Consequences

CEA is insufficiently sensitive to be used alone, even if a low threshold is applied. It is therefore essential to augment CEA monitoring with another diagnostic modality (e.g. CT imaging at 12-18 months) in order to avoid missed cases. Trying to improve sensitivity by adopting a low threshold is a poor strategy because of the high numbers of false alarms generated. We therefore recommend monitoring with more than one diagnostic modality but applying the highest cut-off assessed (10ug/l). It is likely that diagnostic performance could be improved further by treating CEA as a monitoring rather than one-off diagnostic test, and making decisions to investigate further on the basis of trend, but we could not address this issue within the constraints of current Cochrane diagnostic review methodology.

Credits

  • Brian Nicholson, Clinical Biochemistry, Oxford University Hospitals NHS Trust, Oxford, UK
  • Bethany Shinkins, Clinical Biochemistry, Oxford University Hospitals NHS Trust, Oxford, UK
  • Indika Pathiraja, Clinical Biochemistry, Oxford University Hospitals NHS Trust, Oxford, UK
  • Nia Roberts
  • Tim James, Department of Surgery, University of Southampton, Southampton, UK
  • Susan Mallett, Clinical Biochemistry, Oxford University Hospitals NHS Trust, Oxford, UK
  • Rafael Perera, Clinical Biochemistry, Oxford University Hospitals NHS Trust, Oxford, UK
  • John Primrose, Bodleian Health Care Libraries, University of Oxford, Oxford, UK
  • David Mant, Clinical Biochemistry, Oxford University Hospitals NHS Trust, Oxford, UK