Volume 137, Issue 1 , Pages 254-255, January 2009
Remediastinoscopy: A statistical reinterpretation
Article Outline
CTSNet classification: 10
To the Editor:
We read with considerable interest the study by Marra and colleagues1 in the April 2008 issue, “Remediastinoscopy in Restaging of Lung Cancer After Induction Therapy.” We do, however, have some issues with the statistical interpretation of their results.
First, there is the reporting of a P value of 0.0000 in their Table 2.1 A P value is the probability of observing the value or more extreme values given that the null hypothesis is true (ie, no true difference). Marra and colleagues1 give this as 0, which is not possible, when what they mean is that the probability (P value) is less than .0001. They do not state in the table to what test or null hypothesis the P value refers, so it is not possible for the reader to interpret the statistical significance.
Second, Marra and colleagues1 report that remediastinoscopy is associated with 100% specificity. That in itself is to be expected, simply because for the vast majority of tests (endobronchial ultrasonography included) performed with diagnostic biopsy, when tumor is seen in a biopsy specimen, it is invariably confirmed on the reference test (there would be serious problems if tumor diagnosed on the index test proved to be normal on the reference test). Therefore specificity (ability to rule in disease) is not the major focus for surgeons when evaluating remediastinoscopy for restaging the mediastinum. The same specificity will be obtained with any other tests combined with tissue biopsy, such as endobronchial ultrasonography (as referenced by Marra and colleagues1).
Third, what is critical is sensitivity (the ability to rule out disease), so that patients with residual N2 disease are not brought to operation mistakenly, and the reported sensitivity for remediastinoscopy was 61%. It is important for readers to bear in mind that sensitivity truly starts at 50% (a test with no sensitivity, or equipoise), not 0%. Marra and colleagues1 do not report a confidence interval for this estimate; when one is calculated, it ranges from 42% to 77%. That is, it crosses 50%. This low value of sensitivity is entirely in keeping with the previous literature on remediastinoscopy. De Leyn and associates2 attributed this to the technical difficulty in obtaining adequate biopsy specimens from the areas of previous lymph node involvement.
Fourth, Marra and colleagues1 have used their own arbitrary definition of “diagnostic accuracy,” which includes (see footnote in their Table 3) in the denominator “no biopsy.” This does not comply with current Standards for Reporting of Diagnostic Accuracy (STARD).3 The importance of STARD was previously highlighted by Eugene Blackstone4 in this Journal in 2004. We believe that compliance with STARD is crucial for readers (which is why it is part of this Journal's policy), and we are surprised that a non–STARD compliant article was highlighted for continuing medical education credit.
On the basis of these findings, we believe that the conclusion that “remediastinoscopy provides a histologic proof of mediastinal downstaging with high diagnostic accuracy” cannot be sustained by the results presented by Marra and colleagues.1 The results do indicate that residual disease can ruled in with repeated mediastinoscopy, but the procedure is unable to rule out residual mediastinal disease with any degree of certainty (sensitivity of 61%, 95% confidence interval 42%–77%). Futile thoracotomy thus would not be avoided for a large proportion of patients were this technique used to restage the mediastinum after induction chemotherapy.
References
- . Remediastinoscopy in restaging of lung cancer after induction therapy. J Thorac Cardiovasc Surg. 2008;135:843–849
- Prospective comparative study of integrated positron emission tomography–computed tomography scan compared with remediastinoscopy in the assessment of residual mediastinal lymph node disease after induction chemotherapy for mediastinoscopy-proven stage IIIA-N2 non-small-cell lung cancer: a Leuven Lung Cancer Group study. J Clin Oncol. 2006;24:3333–3339
- Toward complete and accurate reporting of studies of diagnostic accuracy. The STARD initiative. Am J Clin Pathol. 2003;119:18–22
- . Caveat emptor: the treachery of work-up bias. J Thorac Cardiovasc Surg. 2004;128:341–344
PII: S0022-5223(08)01505-5
doi:10.1016/j.jtcvs.2008.04.032
© 2009 The American Association for Thoracic Surgery. Published by Elsevier Inc. All rights reserved.
Volume 137, Issue 1 , Pages 254-255, January 2009