Commentary: Coronary revascularization therapies and number needed to treat.

Arnaldo Dimagli, MD (left), Aaron J. Weiss, MD (center), and Faisal G. Bakaeen, MD (right)

Number needed to treat makes the case for the incremental benefit of internal thoracic artery grafting compared with other revascularization strategies, but not without shortcomings.

See Article page 106.

The groundbreaking study from Loop and colleagues associated internal thoracic artery (ITA) bypass to the left anterior descending artery with improved survival compared with all-venous grafting. Indeed, the unique biology of the ITA, its resistance to atherosclerosis, and superior patency has given coronary artery bypass grafting (CABG) a lasting edge over other therapies for coronary artery disease (CAD).

In this issue of the Journal, Dr Ohno reviews the evidence on the long-term survival benefit of CABG with ITA in comparison with medical therapy and percutaneous coronary intervention (PCI) in patients with CAD. He espouses the concept of number needed to treat (NNT), a statistic indicating the number of patients “on average” who need to receive one treatment versus another to prevent the occurrence of an outcome in one patient at a specific time point. He presents the results of seminal studies in coronary revascularization in terms of NNT, which he calculated from the absolute risk reduction, arguing that NNT is a better way to express the magnitude of benefit associated with ITA grafting compared with other CAD therapies. On face value, NNT may appear intuitive and user-friendly. A lower NNT is associated with a more favorable risk–benefit ratio for the treatment under investigation; however, the optimal NNT varies and is dependent on the clinical context.

Moreover, when considering studies reporting time-to-event end points, the calculation of NNT is potentially misleading. This is because analyses of time-to-event outcomes need to account for progressive patient censoring and variation in follow-up. Failure to do so leads to inaccurate estimations of treatment effect magnitude. Furthermore, when survival curves cross or trend close (as in Figure 2 in Ohno's manuscript), alternative calculations have been proposed that obtain more accurate NNT estimates than calculations based solely on the absolute risk reduction.

Dr Ohno postulates a progressive reduction in the NNT due to a progressive increase in benefit of CABG versus PCI over time. From a mechanistic point of view, this makes sense, since CABG addresses the culprit lesions and provides prophylactic surgical collateralization of non–flow-limiting lesions that can prevent future myocardial infarction. In contrast, PCI only treats the culprit lesions and does not provide any prophylactic benefit beyond the stented segment.

The conditio sine qua non for this advantage of CABG is that grafts stay patent. Venous grafts are known to have lower patency rates than arterial grafts. The multiarterial grafting (MAG) strategy is not discussed by Dr Ohno because he focuses his manuscript on the summary of high-quality evidence from randomized clinical trials and the bulk of evidence supporting MAG is from observational studies. The benefit of MAG relies on a delicate balance between patient selection, target vessel choice, surgeon skill, and experience.,

In conclusion, CABG using an ITA is an effective and durable therapy for CAD, and NNT is an imperfect statistical tool to quantify the magnitude of benefit from CABG when compared with alternative therapies.