The STICH trial: Misguided conclusions

STICH Trial Background and Execution 

Initial planning required exclusion of the culprit scar with nonviable regions confirmed by nuclear scans to avoid confusion of hibernation within viable muscle. Two new steps were required. First, participating cardiologists had to measure volume in trial entry patients with ejection fractions of less than 35%, since differing volumes may have a similar ejection fraction (Figure 2, B). Second, surgeons had to incise the normal epicardium over underlying thick-walled scar and reduce volume 30% below baseline levels. This amount of volume reduction was selected because approximately 40% volume reduction (Table 1) was needed for clinical improvement in prior SVR reports of more than 5000 patients. STICH measured volume in only 161 of 490 SVR patients and concluded that SVR is not better than CABG. Unfortunately, no evaluation could be made in 66% of SVR patients whose volume was not measured. CHF severity reached grade III/IV (New York Heart Association) in only 49% of patients in the STICH cohort versus approximately 67% in SVR registry reports,4 but post hoc analysis can make such data-based comparisons.

Table 1. Left ventricular end-systolic volumes before and after surgical ventricular restoration

			ESVI (mL/m2)
	Author(s) (year)	No. of patients	Preop.	Postop.	Reduction (%)
	Di Donato et al8 (2009)
	Type 1	56	83	35	48 (58)
	Type 2	55	87	39	48 (55)
	Type 3	67	96	57	39 (41)
	Suma et al9 (2009)	76	123	74	49 (40)
	Dor et al10 (2008)	104	93	51	42 (45)
	Menicanti et al11 (2007)	301	173∗	100∗	73∗ (42)
	O'Neill et al12 (2006)	135	120	77	43 (36)
	Adams et al13 (2006)	8	92	59	33 (36)
	Schreuder et al14 (2005)	9	92	45	47 (51)
	Tulner et al15 (2006)	21	186∗	101∗	85∗ (46)
	Yamaguchi et al16 (2005)	20	137	65	72 (53)
	Mickleborough et al17 (2004)	41	97	65	32 (33)
	Athanasuleas et al (RESTORE group4) (2004)	671	80.4	56.6	24 (30)
	Jones et al (STICH group1) (2009)	161	83	67	16 (19)

Data compiled by Hisayoshi Suma, MD. ESVI, End-systolic volume index.

∗ End-systolic volume (mL), not indexed.

Two entry criteria existed in the original protocol. First, LV stroke volume index must be beyond 60 mL/m2. The background volume requisite for surgical intervention stemmed from the report by White and associates2 and from experience in aortic and mitral valve disease, where progressive remodeling is not prevented if valve interventions are made after end-systolic volume index (ESVI) exceeds 60 mL/m2 (Figure E1).18 Second, documentation of greater than 35% akinesia from anterior wall necrosis was necessary to show that the culprit scar caused remote muscle dilatation that exceeded the 60 mL/m2 entry point.

These established primary objectives became altered in 2006 (http://clinicaltrials.gov/show/NCT00023595). Such protocol deviation from approved entry criteria is possible by executive decision by the principal investigator, can only occur with approval of the National Institutes of Health grant administration officer, and does not require approval from the other trial participants. Investigational review board notification of this change to other trial participants is their responsibility. The new primary objective included “documented LV anterior wall dysfunction” if necrosis and volume measurements were absent. Moreover, 13% of patients had not had a myocardial infarction.

These primary goal changes introduced wall motion disorders that may occur without necrosis after acute ischemia or hibernation or may follow scar. Ischemic muscle without scar may recover after CABG,19 but this cannot happen if gadolinium magnetic resonance studies show greater than 50% of muscle is scarred.20 SVR has never been recommended in ventricles without scar, and applying SVR in hibernating myocardium was not reported by Dor3 in more than 1000 patients or the 1198 RESTORE patient cohort, in which all patients displayed electrocardiographic and imaging evidence of large akinetic or dyskinetic scar with greater than 60 mL/m2 ESVI measurements.

Echocardiography was used to measure LV volume in the NEJM report but was evaluated in only 38% of STICH patients (212/490 CABG patients and 161/490 CABG and SVR patients). Application of this echo-measuring technique contradicts original and present protocol requirements (Supplemental Data 1), where echocardiography is accepted during screening to measure ejection fraction, but 100% of consenting patients required cardiac magnetic resonance (CMR) or gated single photon emission computed tomography scans to measure volume before and after SVR or CABG. Grant planning excluded 2-dimensional echocardiographic measurement of volume because regional asynergy in ischemic disease makes it unreliable for volume measurement,21 but this limitation becomes offset by CMR 3-dimensional volume analysis.

LV volume was reduced by 19% in SVR patients to reach a volume end point that reflects an inadequate repair as determined by the Surgery Therapy Committee, whose “acceptable STICH procedure” guideline required a 30% ESVI decrease at 4-month CMR measurement (Supplemental Data 2). Failure to achieve benchmark volume reduction differs markedly from SVR values shown in Table 1, averaging 40% reduction and ranging between 30% and 58%. Consequently, the STICH SVR procedure may have reflected a small LV plication or limited intracavitary reconstruction.

Alteration of primary aim changes was suggested in a 2005 trial editorial, “To STICH or not to STICH: We Know the Answer, but Do We Understand the Question,”22 which reached conclusions about questions the grant was assigned to test. Publication of this report led to an immediate editorial in this Journal, “Questions and Answers About the STICH Trial: A Different Perspective,”23 in which answers to original grant application goals were requested.

STICH Flaws 

The credibility of the STICH trial is questioned by flaws addressed in trial background above and execution sections by those listed below. First, all patients needed to have akinesia, yet only 50% displayed this finding. Second, akinesia developed from regional necrosis of 35% of muscle, yet the report fails to document this scar finding (Supplemental Data 3). Third, CMR quantification of ventricular volume is needed in all patients before and after SVR. Instead, 19% underwent an invalid echocardiographic measurement, despite pretrial contact showing CMR measurement capacity in all initial 50 trial centers (Supplemental Data 4). Fourth, 100% required CMR volume measurement for trial entry, yet only 38% (CABG and CABG plus SVR) had any form of volume measurement. Fifth, SVR is indicated only if ESVI is beyond 60 mL/m2, yet volume measurements for SVR or CABG without SVR candidates were not reported. Sixth, 30% reduction of ESVI at 4 months by CMR study is required for acceptable SVR procedure, but ESVI was lowered only 19% in the 33% of patients to demonstrate an inadequate end point. Seventh, the original trial included 50 centers, averaging approximately 10 cases per center. Actually, 96 centers were used, averaging approximately 5 procedures per center, so that Surgical Therapy Committee outcome validation during accreditation becomes a vital earmark before trial enrollment. Its initial meeting outlined 10 cases per center and 5 per surgeon, demonstrating the required 30% volume reduction. Instead, no specification is shown that pretrial ESVI reduction fulfilled trial eligibility criteria.

Evidence Based Medicine and Role of Experience 

The hallmark of technical competence after a surgical procedure requires reaching goals that comply with yardsticks of procedural success. For SVR, favorable clinical results followed approximately 40% volume reduction below control levels in more than 1500 cases in 12 worldwide centers (Table 1). In contrast, STICH reduced ventricular volume 19% in 161 patients, and 96 centers were required to achieve this end point. No volume studies occurred in 66% of patients and an invalid echo-based monitoring method was used in the others. These limitations did not prevent STICH end point interpretation that SVR should be abandoned because CABG achieved similar results. The “evidence based medicine” analysis used to justify this conclusion does not consider how results relate to evaluation of nonapproved volume outcomes in only 33% of patients or how experience influences end point validity.

The 2005 editorial by one of us23 (G.D.B.) suggested potentially inaccurate conclusions after volume reduction by comparing STICH to the 1976 Veterans Administration Hospital CABG study24 of ischemic coronary artery disease revascularization versus medical therapy. The 300 surgically treated patients in 12 centers had 6% mortality (range 3%–12%) and closed conduits related to individual center experience. This contrasts with 1% mortality and superb graft patency in a 1978 report of 1000 patients from the Cleveland Clinic25 to demonstrate how experience based observations influence subsequent management. STICH trial inexperience potential may relate to the learning curve of adequately reducing ventricular volume while doing a new procedure when only about 5 cases per center are performed. Misguided results may camouflage proper outcome analysis and impair correct treatment in the same way that coronary revascularization can be considered unsuccessful when only internal thoracic artery grafting is done without addressing a stenotic right coronary artery, or concluding that antihypertensive drugs are ineffective in studies that used inadequate medication dosages.

We believe that differences between STICH outcomes and worldwide SVR data displaying greater volume reduction indicate that the wrong operation, using the wrong volume measurement monitors, was done on the wrong patients and resulted in the wrong conclusions. This outcome deprives cardiologists of understanding the potential role of volume reduction to treat CHF from ischemic dilated cardiomyopathy and demonstrates that the goals of evidence based medicine were not achieved.

Casualties of STICH Outcomes 

The first casualty of this report may be the increasing CHF population with dilated hearts, which makes up approximately half of the CHF population,26 whereby misguided conclusions may hamper proper treatment decisions. STICH end points do not relate to the SVR procedure that requires properly measuring necrosis and accurate ventricular volume before reducing volume more than 30% below preoperative levels.

Second, STICH may adversely affect cardiac surgery evolution and nontransplant heart failure surgery by limiting development of SVR, a unifying geometric treatment of dilated cardiomyopathy from ischemic and nonischemic causes. Moreover, SVR may link with LV assist devices to introduce a “bridge to restoration” approach and ultimately may create a surgical scaffold that marries this macroscopic approach with cell therapy. The flawed database underlying STICH findings may also limit future government interest in funding other surgical trials.

The third casualty is scientific integrity, because measuring volume rather than ejection fraction as the natural surrogate for development of adverse clinical findings in dilated hearts becomes impaired by STICH findings.

The fourth casualty may be health care cost of CHF treatment, which may exceed $1 trillion per year in 2030 (Figure E2).27

STICH Conclusions and Questions to be Answered 

The following debate questions still require a proper response. First, is ESVI measurement important in evaluating and treating dilated cardiomyopathy? Second, why were incorrect volume measurements used? Third, does CABG treatment lead to improved function in scarred muscle? CABG is used in the non-SVR group that does not exclude the scarred segment. Fourth, where are viability measurements showing 35% regional necrosis in all included patients? Fifth, is there confirmed validation of center–surgeon eligibility showing greater than 30% volume reduction to comply with Surgical Therapy Committee guidelines? Sixth, did STICH study ischemia (with or without hibernation) or postmyocardial infarction anterior scar? Seventh, why were the primary objectives changed? Eighth, do STICH conclusions invalidate the extensive SVR patient registry database?

STICH Conclusions and Meaningful End Points 

This analysis of STICH report flaws and conclusions suggests the trial should retract the NEJM report or reanalyze data to achieve meaningful results. The target goal must (1) exclude all patients with invalid echocardiographic volume measurements, (2) include only patients in whom regional nonviability of greater than 35% akinesia is documented by nuclear medicine scans, (3) quantify all patients with greater than 30% volume reduction by CMR study, and (4) report only patients with “acceptable” volume reduction by CMR at 4 months. The STICH trial must address each of these questions because misguided STICH conclusions contradict the role of augmented ventricular volume as the surrogate for the natural history of increasing morbidity and mortality in dilated hearts.2, 18 Without this action, the STICH trial conclusions simply show that statisticians can defy nature from a flawed database.