In 2007, the Lancet published the results of the Valsartan in a Japanese population with hypertension and other cardiovascular disease (Jikei Heart Study): a randomised, open-label, blinded endpoint morbidity-mortality study (1). According to the authors, the addition of the angiotensin receptor blocker (ARB) valsartan to standard cardiovascular treatment, compared with an increased dose or number of standard non ARB drugs, reduced the incidence of the primary composite endpoint, of heart, brain, and kidney complications (1). The main effect of addition of valsartan was mainly driven by a reduction in stroke, angina pectoris, dissecting aortic aneurysm, and heart failure incidence rate (1). Furthermore, the authors suggested that since patients in both treatment groups showed a similar degree of blood pressure control, the benefit observed with the ARB could not entirely be explained by differences in blood pressure (1,2). Having in mind that Asian patients have often been under-represented in cardiovascular trials, the authors were initially congratulated on their report, mainly because of conducting a randomized trial in Japanese patients with cardiovascular disease (2).
However, simultaneously with the Jikei Heart Study publication the accompanying Lancet editorial underscored several major weaknesses of this clinical trial (2). The commentators highlighted the inadequate sample size, the lower incidence of “hard” end-points compared to other similar trials as well as the insufficient dosing both for valsartan and for non-ARB therapy (2). Moreover, it was acknowledged that the primary composite end-point was mainly driven by reductions in new admissions for stable angina and heart failure rather than stroke or myocardial infarction as one would commonly expect (2). The use of an open-blinded-label end-point design as well as the inclusion of transient ischemic attacks – a “weak” not commonly used in clinical trials event- as an end-point were also criticized (2). More importantly, the commentators raised concerns regarding the statement of the authors that “Blood pressure and heart rate did not differ between the valsartan regimen and the control regimen throughout the trial (table 3, p=0•196 for systolic blood pressure and p=0•176 for diastolic blood pressure at end of study)”. In specific, they provided data based on large sample Z-tests showing that blood pressure levels reported by the investigators were in-fact significantly different between treatment groups both at baseline and during follow up. This major concern alongside with observations from other studies showing that the immediate blood pressure lowering translates into early clinical benefit and that even small gradients in the achieved blood pressure levels explains most of the differences in cardiovascular outcomes, led the commentators to question the suggestion made by the investigators that blood pressure was not a major determinant of outcomes (2). Staessen and Richart concluded that “Nevertheless, one should not accept at face value the main conclusions of the Jikei report” (2).
Following the publication of the study two correspondence letters received and published by the Lancet raised concerns regarding the integrity of the data (3,4). First in 2007, Boutitie et al, commented the controversial role of a 2.1 mmHg lower systolic blood pressure at 6 months observed with valsartan use on the final outcome (3). This time the commentators showed that the addition of the observed benefit in stroke rate with valsartan in the Jikei Heart Study (odds ratio of 0.60 for stroke associated with a -2.1 mmHg reduction in systolic blood pressure) to a figure of the meta-analysis by Staessen and colleagues (depicting the relation between observed odds ratios for fatal and non-fatal stroke and corresponding differences in systolic blood pressure) resulted in situating Jikei trial someway below the lower limit of the 95% confidence interval of the regression line (3). In addition, Boutitie at al, highlighted the discordance between the Jikei’s reported association (stroke risk reduction of 38.7% associated with a -2.1 mmHg reduction in systolic blood pressure) and the association (26% stroke risk reduction associated with a -5 mmHg reduction is systolic blood pressure) reported by the Blood Pressure Lowering Treatment Trialists’ collaboration (3). Second in 2012, Yui reported a coincidence of identical mean and standard deviation of achieved systolic blood pressure between the valsartan and the comparator group, an observed quality that was not present at baseline (4). Therefore, Yui raised a concern of how an initially randomized and heterogeneous population became homogeneous after a 3-year drug observation (4). Furthermore, Yui identified several odd coincidences between Jikei and Kyoto Heart study regarding the reported blood pressure means and standard deviations (4).
All this accumulated concerns as well as the retraction of the Kyoto Heart Study by the European Heart Journal in February 2013, in which several authors were in common with the Jikei Heart Study and both studies investigated valsartan, led to an investigation into the conduct of the Jikei Heart Study (5). The report identified concerns over the reliability of the blood pressure data and the suspected data alteration during statistical analysis (5). These findings led to sufficient doubt as to the integrity of the Jikei Heart Study and finally to the formal retraction of the paper from the scientific record.
According to our point of view, this damaging “story” for the medical community is very important for several educational and audit-wise research issues. Tanimoto and colleagues in a letter to the Lancet editors in 2013 elegantly summarized the lessons learned from data falsification in clinical trials (6). Firstly, it has been shown that scientific fraud can pass through even the most meticulous peer-review processes (6). Secondly, that open discussions among the medical community as reported by journal sections such as Letter to the editors or Correspondence columns play a major role in expressing criticism, in improving medical journals and also in identifying research misconduct (6). Finally, that the Committee on Publication Ethics is now more than ever needed in advocating, developing and organizing systems that can prevent and identify misconduct (6).
Critical and continuous review and discussion even after publication is a prerequisite before dissemination of medical information to the public or implementation in clinical practice. More importantly we should not forget that as practicing physicians we ought to criticize all novel medical information in context with our everyday clinical experience, our common medical sense and our personal observations.
References
1. Mochizuki S, Dahlöf B, Shimizu M, Ikewaki K, Yoshikawa M, Taniguchi I, Ohta M, Yamada T, Ogawa K, Kanae K, Kawai M, Seki S, Okazaki F, Taniguchi M, Yoshida S, Tajima N; Jikei Heart Study group. Valsartan in a Japanese population with hypertension and other cardiovascular disease (Jikei Heart Study): a randomised, open-label, blinded endpoint morbidity-mortality study. Lancet. 2007;369:1431-9.
2. Staessen JA, Richart T. Sum and substance in the Jikei Heart Study. Lancet. 2007;28:369:1407-8.
3. Boutitie F, Gueyffier F, Achard JM, Temmar M, Fournier A. The JIKEI trial. Lancet. 2007;370:1825-6.
4. Yui Y. Concerns about the Jikei Heart Study Lancet. 2012;379:e48.
5. The Lancet Editors. Retraction—Valsartan in a Japanese population with hypertension and other cardiovascular disease (Jikei Heart Study): a randomised, open-label, blinded endpoint morbidity-mortality study. Lancet 2013;382: 843
6. Tanimoto T, Kami M, Shibuya K. Research misconduct and scientific integrity: a call for a global forum.Lancet. 2013;382:940.
Citation: