**6. Factors favoring surgery as the mode of revascularization**

### **6.1. LV dysfunction**

LV function has never been shown to have a significant interaction with mode of revasculari‐ zation (PCI versus CABG) with regard to survival. In fact, the majority of studies comparing PCI with CABG enrolled a low percentage of patients with abnormal LV function *(20 percent or less)* [45]. However, it has been considered an important variable that favors revasculariza‐ tion with surgery due to historical data showing that patients with significant LV dysfunction have improved survival with CABG compared to medical therapy. An initial signal for preferential benefit of revascularization in patients with mild to moderate LV dysfunction (*LV EF of 35-49%*) was seen in subgroup analyses of the CASS randomized study and the VA study [46, 47]. This was further supported by a meta-analysis demonstrating a significantly longer survival time in 10-year follow up with surgical revascularization over medical therapy in patients with LV dysfunction [10.6 months] compared with those with normal LV function [2.3 months] [27]. These studies however did not address whether a similar effect would be seen in severe LV dysfunction.

in non-diabetics there was no overall benefit of surgery over angioplasty[45]. This relationship was corroborated by significant interaction found for the diabetic subgroup (p=0.014) [45]. The interaction was robust and was present even after excluding the data contributed by BARI and when adjusted for various clinical parameters (age, sex, smoking, hypertension, history of MI,

Multivessel Disease in the Modern Era of Percutaneous Coronary Intervention

http://dx.doi.org/10.5772/55103

325

One hypothesis that may explain why diabetics may have a better outcome with surgery than angioplasty is that restenosis may be more aggressive in this group of patients [20]. It has been well established that one major drawback of percutaneous coronary intervention is the need for repeat revascularization due to restenosis [20]. While restenosis itself may not incur an in‐ creasedriskofmortality,therepeatexposuretotheinherentriskofinterventionmaybeadditive. TheneedforrepeatrevascularizationwascertainlymorestrikingintheearliertrialswherePOBA was used (54 percent within 5 years for the BARI trial) [50]. There has been considerable improvement to both the techniques and technology of PCI first with the development of BMS andnowDESwhichare intendedto reducedrestenosis rateshavebeenintegratedinto common practice. However, eveninmore contemporary trials suchas *Synergy betweenPCIwithTaxus and Cardiac Surgery (*SYNTAX), DES conferred a 13.5% need for revascularization compared with

*BARI 2D* is a contemporary trial evaluating revascularization (both PCI and CABG) with intensive medical therapy for SIHD in diabetic patients. No significant difference emerged overall between the two groups in terms of overall mortality, cardiac death or myocardial infarction over the 5-year follow up period [52]. The mode of revascularization was at the discretion of the treating physician and the burden of disease tended to be higher in the CABG group than the PCI group with a mean number of lesions being 5.6 versus 4.3 respectively [52]. Although not specifically designed to compare PCI and CABG as a mode of revascularization, it is still noteworthy that there was a significant difference in death and MI for those revascu‐ larized with CABG compared to medical therapy (21.1% versus 29.2% p=0.010); the same

To date, the trial most relevant in determining whether the difference between PCI and CABG for revascularizing SIHD in diabetic patients in a randomized fashion is the *FREEDOM* trial [53]. There are 1900 diabetic patients with multivessel disease (defined by >70% stenosis in 2 or more epicardial vessels supplying different vascular territories) that were randomized to CABG versus PCI with DES (Paclitaxel or Sirolimus eluting stent at the discretion of treating physician) with a background of guideline supported optimal medical therapy. The primary endpoint is a composite of all-cause mortality, non-fatal MI or stroke over the mean follow up period of 4.37 years [53]. This consists of a high-risk population with 83% having 3-vessel disease and a significant proportion (32%) requiring insulin therapy [53]. The study results showed that there was a reduction in *primary endpoint all cause death, non-fatal MI and stroke in the CABG group,* with an **ARR of 7.9% (**26.6% in the PCI group and 18.7% in the CABG group, **p=0.005) and a Number Needed to Treat (NNT) of 12.5 [Figure 2] [53]**. There was also a reduction in *all-cause mortality* which was 16.3% in the PCI arm and 10.9% in the CABG arm with an **ARR = 5.4% and NNT of 19 [Figure 2] [53]**. This was at the cost of an increase in stroke, as might be expected in the CABG arm of **2.8% Number Needed to Harm (NNH) of 36 [53]**.

5.9% in patients treated with CABG over a one-year period [51].

comparison in the PCI stratum revealed no difference [52].

heart failure and 3VD) [45].

The Surgical Treatment for Ischemic Heart Failure (STITCH) trial published recently in 2011, designed to address this question in a randomized comparison of medical therapy versus surgical revascularization in patients with an *EF of 35 percent or less [48]*. There was a nonstatistically significant trend (p=0.12) towards decreased all-cause mortality in the surgical group with an relative reduction of 24% and an ARR of 5 percent over the six year follow up period [48]. The lack of statistical significance, in the context of intention to treat analysis, may be related to the disproportionate crossover rate with 17 percent of the patients assigned to medical therapy ultimately receiving coronary bypass surgery [48]. Nevertheless, there was still a significant relative reduction of death from cardiovascular causes of 19 percent (p=0.05) and a significant relative reduction in death from any cause and hospitalization from cardio‐ vascular causes of 26 percent (p<0.001)[48].

It is unclear why surgical revascularization confers clear benefit in mild to moderate LV dysfunction, and only modest benefit severe LV dysfunction, but there are possible explana‐ tions. Medical therapy has advanced tremendously since the initial comparisons between medical therapy and surgery, which may decrease the relative mortality benefit between the two treatments during this more contemporary comparison. One could also hypothesize that the beneficial effect of revascularization plateaus at the extremes of LV dysfunction due to irreversible remodelling and/or the progressive increase in associated procedural risk.

There is currently limited data in how revascularization with PCI would affect prognosis in the setting of LV dysfunction[14]. As a result, the most recent ACC/AHA guidelines still recommend CABG for patients with LV EF 35-50 percent with a *IIa recommendation (Grade B evidence)* and a *IIb recommendation (Grade B evidence)* for those with a LV EF of less than 35 percent [49]. And currently, the ACC/AHA guidelines state that there is *insufficient data* to make a recommendation for revascularization with PCI in patients with LV dysfunction [49]. In practice however, if a patient does require revascularization but is not a surgical candidate, that natural decision is that if percutaneous intervention is feasible, that option should be entertained.

#### **6.2. Diabetes favoring surgical revascularization**

The Bypass Angioplasty Revascularization Investigation (BARI) was one of the first major landmark studies comparing angioplasty versus coronary bypass in patients with both stable anginal symptoms and unstable angina with MVD. Although there was no significant difference between the two treatments in the overall survival, patients with diabetes tended to have a significantly lower mortality rate with CABG (19.4%) compared with angioplasty (34.5%) with an **ARR of 15.1%** [50]. This finding was further confirmed with a meta-analysis comparing angioplasty with CABG in MVD, albeit with a smaller ARR of 7.7% [45]. In contrast, in non-diabetics there was no overall benefit of surgery over angioplasty[45]. This relationship was corroborated by significant interaction found for the diabetic subgroup (p=0.014) [45]. The interaction was robust and was present even after excluding the data contributed by BARI and when adjusted for various clinical parameters (age, sex, smoking, hypertension, history of MI, heart failure and 3VD) [45].

have improved survival with CABG compared to medical therapy. An initial signal for preferential benefit of revascularization in patients with mild to moderate LV dysfunction (*LV EF of 35-49%*) was seen in subgroup analyses of the CASS randomized study and the VA study [46, 47]. This was further supported by a meta-analysis demonstrating a significantly longer survival time in 10-year follow up with surgical revascularization over medical therapy in patients with LV dysfunction [10.6 months] compared with those with normal LV function [2.3 months] [27]. These studies however did not address whether a similar effect would be

The Surgical Treatment for Ischemic Heart Failure (STITCH) trial published recently in 2011, designed to address this question in a randomized comparison of medical therapy versus surgical revascularization in patients with an *EF of 35 percent or less [48]*. There was a nonstatistically significant trend (p=0.12) towards decreased all-cause mortality in the surgical group with an relative reduction of 24% and an ARR of 5 percent over the six year follow up period [48]. The lack of statistical significance, in the context of intention to treat analysis, may be related to the disproportionate crossover rate with 17 percent of the patients assigned to medical therapy ultimately receiving coronary bypass surgery [48]. Nevertheless, there was still a significant relative reduction of death from cardiovascular causes of 19 percent (p=0.05) and a significant relative reduction in death from any cause and hospitalization from cardio‐

It is unclear why surgical revascularization confers clear benefit in mild to moderate LV dysfunction, and only modest benefit severe LV dysfunction, but there are possible explana‐ tions. Medical therapy has advanced tremendously since the initial comparisons between medical therapy and surgery, which may decrease the relative mortality benefit between the two treatments during this more contemporary comparison. One could also hypothesize that the beneficial effect of revascularization plateaus at the extremes of LV dysfunction due to irreversible remodelling and/or the progressive increase in associated procedural risk.

There is currently limited data in how revascularization with PCI would affect prognosis in the setting of LV dysfunction[14]. As a result, the most recent ACC/AHA guidelines still recommend CABG for patients with LV EF 35-50 percent with a *IIa recommendation (Grade B evidence)* and a *IIb recommendation (Grade B evidence)* for those with a LV EF of less than 35 percent [49]. And currently, the ACC/AHA guidelines state that there is *insufficient data* to make a recommendation for revascularization with PCI in patients with LV dysfunction [49]. In practice however, if a patient does require revascularization but is not a surgical candidate, that natural decision is that if percutaneous intervention is feasible, that option should be

The Bypass Angioplasty Revascularization Investigation (BARI) was one of the first major landmark studies comparing angioplasty versus coronary bypass in patients with both stable anginal symptoms and unstable angina with MVD. Although there was no significant difference between the two treatments in the overall survival, patients with diabetes tended to have a significantly lower mortality rate with CABG (19.4%) compared with angioplasty (34.5%) with an **ARR of 15.1%** [50]. This finding was further confirmed with a meta-analysis comparing angioplasty with CABG in MVD, albeit with a smaller ARR of 7.7% [45]. In contrast,

seen in severe LV dysfunction.

324 Artery Bypass

vascular causes of 26 percent (p<0.001)[48].

**6.2. Diabetes favoring surgical revascularization**

entertained.

One hypothesis that may explain why diabetics may have a better outcome with surgery than angioplasty is that restenosis may be more aggressive in this group of patients [20]. It has been well established that one major drawback of percutaneous coronary intervention is the need for repeat revascularization due to restenosis [20]. While restenosis itself may not incur an in‐ creasedriskofmortality,therepeatexposuretotheinherentriskofinterventionmaybeadditive. TheneedforrepeatrevascularizationwascertainlymorestrikingintheearliertrialswherePOBA was used (54 percent within 5 years for the BARI trial) [50]. There has been considerable improvement to both the techniques and technology of PCI first with the development of BMS andnowDESwhichare intendedto reducedrestenosis rateshavebeenintegratedinto common practice. However, eveninmore contemporary trials suchas *Synergy betweenPCIwithTaxus and Cardiac Surgery (*SYNTAX), DES conferred a 13.5% need for revascularization compared with 5.9% in patients treated with CABG over a one-year period [51].

*BARI 2D* is a contemporary trial evaluating revascularization (both PCI and CABG) with intensive medical therapy for SIHD in diabetic patients. No significant difference emerged overall between the two groups in terms of overall mortality, cardiac death or myocardial infarction over the 5-year follow up period [52]. The mode of revascularization was at the discretion of the treating physician and the burden of disease tended to be higher in the CABG group than the PCI group with a mean number of lesions being 5.6 versus 4.3 respectively [52]. Although not specifically designed to compare PCI and CABG as a mode of revascularization, it is still noteworthy that there was a significant difference in death and MI for those revascu‐ larized with CABG compared to medical therapy (21.1% versus 29.2% p=0.010); the same comparison in the PCI stratum revealed no difference [52].

To date, the trial most relevant in determining whether the difference between PCI and CABG for revascularizing SIHD in diabetic patients in a randomized fashion is the *FREEDOM* trial [53]. There are 1900 diabetic patients with multivessel disease (defined by >70% stenosis in 2 or more epicardial vessels supplying different vascular territories) that were randomized to CABG versus PCI with DES (Paclitaxel or Sirolimus eluting stent at the discretion of treating physician) with a background of guideline supported optimal medical therapy. The primary endpoint is a composite of all-cause mortality, non-fatal MI or stroke over the mean follow up period of 4.37 years [53]. This consists of a high-risk population with 83% having 3-vessel disease and a significant proportion (32%) requiring insulin therapy [53]. The study results showed that there was a reduction in *primary endpoint all cause death, non-fatal MI and stroke in the CABG group,* with an **ARR of 7.9% (**26.6% in the PCI group and 18.7% in the CABG group, **p=0.005) and a Number Needed to Treat (NNT) of 12.5 [Figure 2] [53]**. There was also a reduction in *all-cause mortality* which was 16.3% in the PCI arm and 10.9% in the CABG arm with an **ARR = 5.4% and NNT of 19 [Figure 2] [53]**. This was at the cost of an increase in stroke, as might be expected in the CABG arm of **2.8% Number Needed to Harm (NNH) of 36 [53]**.

**6.3. Degree of ischemia and revascularization: Is the effect independent of symptoms?**

[Figure 3] [54].

Although there is paucity of randomized data addressing the question of how the degree of functional ischemia relates to the benefit of revascularization, there is observational data that suggests a strong relationship [54, 55]. Adjusted risk models have suggested that patients with *less than a 10-12.5% threshold* of ischemia as demonstrated by stress myocardial perfusion imaging, the survival profile of those treated with medical therapy were similar or even perhaps slightly better than those who are treated with revascularization [54]. However *above the threshold of 10-12.5% of ischemia,* there was a graded incremental survival benefit of revascularization over medical therapy, with a risk adjusted relative risk reduction of 50%

Multivessel Disease in the Modern Era of Percutaneous Coronary Intervention

http://dx.doi.org/10.5772/55103

327

**Figure 3.** Log hazard ratio for revascularization (Revasc) versus medical therapy (Medical Rx) as a function of % myo‐ cardium ischemia based on final cox proportional hazards model. Model, P< 0.0001; interaction, P=0.0305 (Repro‐

This data was further corroborated by another observational study demonstrating a similar effect in a group of asymptomatic diabetic patients [55]. This study found a *14% survival benefit* in patients with a high-risk myocardial perfusion scan treated with CABG over medical therapy [55]. Patients treated with PCI in the high-risk scan group did not achieve a survival benefit over medical therapy [55]. This may be due to the fact that, this treatment group consisted of only 10.7% three vessel disease and no patients with left main disease whereas in the CABG group, this was 52.1% and 20.8% respectively [55]. The other caveat to this data is

duced with permission from Hachamovitch et al. Circulation 2003. 107: 2900-2906).

**Figure 2.** Kaplan-Meier Curves for composite primary outcome (all cause death, non-fatal myocardial infarction or stroke) and all cause death in a comparison between PCI with DES compared with CABG for multivessel disease in diabetic patients. Reproduced with permission from Farkough ME et al. NEJM 2012. DOI 10.1056/NEJMoa1211585.

### **6.3. Degree of ischemia and revascularization: Is the effect independent of symptoms?**

Although there is paucity of randomized data addressing the question of how the degree of functional ischemia relates to the benefit of revascularization, there is observational data that suggests a strong relationship [54, 55]. Adjusted risk models have suggested that patients with *less than a 10-12.5% threshold* of ischemia as demonstrated by stress myocardial perfusion imaging, the survival profile of those treated with medical therapy were similar or even perhaps slightly better than those who are treated with revascularization [54]. However *above the threshold of 10-12.5% of ischemia,* there was a graded incremental survival benefit of revascularization over medical therapy, with a risk adjusted relative risk reduction of 50% [Figure 3] [54].

**Figure 3.** Log hazard ratio for revascularization (Revasc) versus medical therapy (Medical Rx) as a function of % myo‐ cardium ischemia based on final cox proportional hazards model. Model, P< 0.0001; interaction, P=0.0305 (Repro‐ duced with permission from Hachamovitch et al. Circulation 2003. 107: 2900-2906).

This data was further corroborated by another observational study demonstrating a similar effect in a group of asymptomatic diabetic patients [55]. This study found a *14% survival benefit* in patients with a high-risk myocardial perfusion scan treated with CABG over medical therapy [55]. Patients treated with PCI in the high-risk scan group did not achieve a survival benefit over medical therapy [55]. This may be due to the fact that, this treatment group consisted of only 10.7% three vessel disease and no patients with left main disease whereas in the CABG group, this was 52.1% and 20.8% respectively [55]. The other caveat to this data is

**Figure 2.** Kaplan-Meier Curves for composite primary outcome (all cause death, non-fatal myocardial infarction or stroke) and all cause death in a comparison between PCI with DES compared with CABG for multivessel disease in diabetic patients. Reproduced with permission from Farkough ME et al. NEJM 2012. DOI 10.1056/NEJMoa1211585.

326 Artery Bypass

the low overall use of optimal medical therapy, which may overestimate the effect size of revascularization in some anatomic subgroups. The use of ASA, ACEi's and BB's were all 40% or less with no mention of statin therapy [55].

In summary, patients with moderate to high-risk scans by myocardial perfusion have signif‐ icant survival advantage if treated with revascularization over medical therapy alone *even in the absence of symptoms* [54, 55]. This benefit has more convincingly been demonstrated in patients undergoing CABG [54, 55]. Although the current standard and use of optimal medical therapy has improved over time, this effect is likely still significant. This survival benefit in the high-risk scans has not been clearly demonstrated in patients treated with PCI in stable coronary disease. Recently, the COURAGE trial involving over 2000 patients, which compared optimal medical therapy plus PCI with optimal medical therapy alone in stable CAD included a high prevalence of MVD (over two thirds) associated with the same proportion of multiple reversible defects by myocardial perfusion imaging [56]. This study showed, after exclusion of patients with high-risk anatomy (LM) and markedly positive exercise stress testing (sub‐ stantial ST depression or hypotensive response in stage I Bruce protocol), no significant difference in all cause death and non-fatal myocardial infarction between the medical therapy and PCI groups [56].

#### **6.4. Impact of complexity of disease**
