Re: Stating the Obvious, for What It’s Worth

Editor:

We read Dr. Murphy’s commentary (1) with interest and acknowledge several points that were raised. Our study (2) retrospectively reviewed 101 renal artery stents treated for in-stent restenosis (ISR). Patients in our study were treated with percutaneous transluminal angioplasty (PTA) or repeat stent implantation. The patients who presented earlier in the study period were generally treated with PTA. Patients who presented later tended to be treated with bare metal stents (BMSs). Only six patients were treated with drug-eluting stents (DESs) toward the end of the study period as the technology became available. Small studies have demonstrated lower rates of renal artery restenosis with the use of DESs, but there is concern that the size of the stents and radial force may limit their use (3,4). However, the populations in those studies were small, and larger prospective trials are needed. Newer therapies such as drug-eluting balloons and endovascular brachytherapy may have a role to play in the management of ISR (5), but were not evaluated in our study (2).

With only 13 subjects who exhibited recurrent ISR requiring a third procedure, all analyses were univariate, as mentioned in our Materials and Methods section. Our 6.89 hazard ratio (HR) of restenosis among BMS-treated arteries uses PTA as the reference. The HR of 6.3 mentioned in the commentary (1) does not take into account the DES group.

We are concerned by the assertions regarding our follow-up duration. The commentary (1) states that patients treated for ISR with a BMS (range, 3–119 mo) were followed much longer than patients treated with PTA (range, 3–15 mo). We respectively disagree with this comment. In fact, these time periods were not reported in our published manuscript (2). It is interesting that the commentary was not based on the manuscript in press. The time periods Dr. Murphy cites were the durations of patency among primary stents in which a second restenosis event developed after initial ISR treatment. We did not report these numbers in our final manuscript because they added little to our conclusion at the expense of potential confusion, as exemplified by the commentary.

As we previously noted, follow-up did not occur at regular intervals. This, along with nonstandardized treatment, is a major limitation of retrospective research (6). Unforeseen biases are inevitable even though we take great strides to minimize their impact. We believe the concerns regarding our Kaplan–Meier curves are based on a previously submitted draft and not the document in press, which had been vetted by peer review. The Kaplan–Meier curves that were published demonstrate the number at risk at each 1-year interval over a period of 5 years (2).

We agree with Dr. Murphy’s statement (1) that retrospective studies such as ours are useful for planning prospective research. However, clinical decision-making should be based on prospective controlled studies when possible. Finally, a power analysis should not be performed in a retrospective analysis and needs to be done as part of a prospective trial and interpreted in the proper context with the use of confidence intervals (7).