Download this article's Factoid (PDF & PPT for Gold Subscribers)

In older patients with multivessel disease, coronary artery bypass graft (CABG) surgery is associated with decreased long-term mortality compared with percutaneous coronary intervention (PCI), according to a large observational study published online April 23, 2013, ahead of print in the Annals of Internal Medicine. The CABG advantage is strongest in patients with certain clinical characteristics, such as diabetes and heart failure, while it is not present or reversed in healthier patients, suggesting the need for physicians to personalize treatment recommendations, the authors say.

Investigators led by Mark A. Hlatky, MD, of Stanford University School of Medicine (Stanford, CA), looked at data from 251,553 Medicare patients who underwent PCI (n = 57,330) or CABG (n = 194,223) for multivessel disease between 1992 and 2008. From this cohort, they analyzed 52,578 propensity-score matched pairs of PCI and CABG patients.

Median follow-up was 4.3 years. Kaplan-Meier analysis estimated 5-year survival to be 74.1% after CABG and 71.9% after PCI, yielding a reduced risk of all-cause mortality with CABG vs. PCI (HR 0.92; 95% CI 0.90-0.95; P < 0.001).

Clinical Characteristics Modify Treatment Effect

Even after additional adjustment for baseline characteristics, the relative mortality benefit of CABG over PCI was accentuated by white race, diabetes, tobacco use, PAD, and heart failure (table 1).

Table 1. Effect of Patient Characteristics on 5-Year Mortality: CABG vs. PCI

^a For interaction.

^b Mainly due to “white” vs. “other race” (P = 0.025); “white” vs. “black” did not reach significance (P = 0.057).

The mean increase in life expectancy from receiving CABG rather than PCI was 0.053 life years (P < 0.001). However, the estimated life years added by CABG varied widely across the study population. Although patients with diabetes, heart failure, tobacco use, or PAD benefited the most from surgery, those lacking these characteristics or a primary diagnosis of MI on hospitalization—41% of the overall cohort—were predicted to have better survival after PCI than after CABG.

The authors observe that due to narrow eligibility requirements and limited sample sizes, randomized trials only show the impact of treatments on the ‘average’ patient. In contrast, the current study makes clear that the comparative effectiveness of CABG vs. PCI differs among clinical subgroups, helping physicians in real-world practice to individualize treatment decisions.

Impact of ‘Unmeasured Confounders’ Debatable

But in a telephone interview with TCTMD, Ajay J. Kirtane, MD, SM, of Columbia University Medical Center (New York, NY), stressed the near impossibility of adequately compensating for unmeasured confounders when detailed clinical information is lacking, as in the current study based on administrative data.

“The truth of all these comparative-effectiveness studies is that they absolutely do not ask all of the questions that we ask as clinicians when we treat patients,” Dr. Kirtane said. “And by not asking all those questions, they are inherently limited in their ability to parse out treatment differences. If there were a way to capture patient frailty or inoperability, for example, then I think we might start believing these data a little more [because] it would make for a fairer comparison.”

But Fred H. Edwards, MD, of the Society of Thoracic Surgeons (Chicago, IL), professed greater confidence in the study’s methods and conclusions. “The propensity matching and other statistical techniques used to mitigate the possibility of unmeasured confounders or selection bias appear to be pretty effective,” he told TCTMD in a telephone interview. “The reason I think we can say that is that the results of both this and the ASCERT trial are quite consistent with what we’re seeing in randomized trials.”

Another drawback of the current study, Dr. Kirtane noted, is that many patients were treated before the DES era. “To conclude from this type of study that one [strategy] is definitively better than the other would be a stretch,” he commented.

While admitting that “it would have been more palatable to some people to focus on the current [stent] era,” Dr. Edwards argued that surgery holds an inherent advantage over stenting, regardless of advances in technology, because bypass grafting addresses both current and future blockages of the target vessel, whereas PCI treats only the current stenosis.

Drs. Edwards and Kirtane agreed that the most important message of the study is that the presence of certain clinical characteristics modifies the relative efficacy of surgery, requiring that treatment recommendations be individualized.

Combination of Anatomic, Clinical Data Most Informative

The 2 commentators were also in accord that the case for surgery is strongest among patients with diabetes, PAD, or heart failure, in whom CAD tends to be more extensive and diffuse. But the absence of a Syntax score characterizing lesion complexity is an important limitation, Dr. Kirtane said. For example, he noted, when diabetes is not linked to diffuse disease, the condition loses prognostic value.

While acknowledging that in the SYNTAX trial a low Syntax score eliminated surgery’s advantage over PCI, Dr. Edwards said that finding “has not been borne out in some subsequent studies.” However, the new Syntax score, which combines anatomic and clinical information, offers more promise for guiding decision making, he said. In the clinical category, patient frailty is widely recognized to be an important factor, he added, noting that the trait will be assessed in a future ASCERT study.

Dr. Edwards said he hoped the concordant results of ASCERT and the current study would encourage greater use of a heart team pairing a surgeon with a cardiologist to assess individual patients and provide them with tailored information on the short- and long-term risks of each procedure. That “would be a huge step forward,” he said.

Study Details

Patients were matched for baseline characteristics as well as multiple medical conditions; they were also matched by year of index procedure, diabetes status, and age within 1 year.

Related Stories:

• ASCERT: Long-term Results Favor CABG Over PCI for Multivessel Disease
• Study Examines Influence of Baseline Factors on Outcomes After CABG, PCI
• FREEDOM: CABG Superior to PCI in Diabetic Patients with Multivessel Disease