Restated Issues and Responses by Bartunek et al. to Issues Raised by Francis et al.
The Appendix includes information on the C-CURE trial (1) in response to questions raised by Francis et al. (2).
1. How many patients were actually randomized? Was it 47 (Figure 1) or 48 (text page 2330) or 45 (baseline characteristics in Table 1)?
Informed consent was requested from 48 patients. One patient refused participation, and of the remaining 47 randomized subjects, two patients were subsequently excluded due to exclusion criteria identified prior to bone marrow harvest, i.e., one patient was diagnosed with a left ventricular aneurysm, a second patient developed a left ventricular thrombus. Table 1 provides baseline characteristics for these 45 patients. There was an imprecision in the text (page 2330) where 48 patients who entered the consent process are described as ‘randomized’ (1).It would have been more precise to state, “Consent was requested from 48 patients, and 47 were randomized…”. Figure 1 depicts the flow of the study, presenting its stepwise execution (1).
2. According to EudraCT (3), the primary endpoint is radionuclide ejection fraction. There seems to be no mention of the results, nor indeed of it ever being carried out in any patient, in the JACC paper. Were the primary endpoint radionuclide EFs done in some or all patients, and not published? If so, why? Or was it decided not to do them in anyone? If so, when?
Study design included a safety/feasibility phase (Stage A; n=45 patients) to be followed by an efficacy phase with a primary endpoint of radionuclide ejection fraction (Stage B; n = 195 patients). Stage A had a built-in go/no go decision at the 3-month follow-up visit. At the 3-month time-point, the procedure was deemed feasible and safe. Yet, with advice received from regulatory authorities that a properly powered study assessing efficacy in heart failure should stand alone and would require inclusion of outcomeendpoints beyond a focus on cardiac function, the Steering Committee communicated to study sites the decision not to proceed with Stage B, and limited the C-CURE trial to a safety/feasibility study completing Stage A per protocol.
3. The JACC paper describes an unrelated primary endpoint for this study. When was the primary endpoint changed? Did all investigators agree?
Feasibility and safety were the primary end-points in the C-CURE study as reported in the manuscript (1). Selection of these primary end-points is standard practice for first-in-man studies assessing stem cell-based therapeutics. As communicated to study investigators across all study sites, the Steering Committee limited the C-CURE trial to a safety and feasibility study.
4. According to Figure 1, 15 were randomized to the control arm. How can those 15 patients consist of 22 males and 2 females, as stated in Table 1?
Protocol defined analysis where all patients not receiving cell therapy were placed in the Control arm (n=24) and all those receiving cell therapy placed in the Cell Therapy arm (n = 21) is presented in Online Figure 13 with patient demographics shown in Table 1 (1) The manuscript also presents in main figures a 7 complementary analysis, as advised in the peer-review process, that maintains randomization with analysis taking into account clinical inclusion/cell release criteria (n = 15 Control; n = 21 Cell Therapy) (1).
5. How can the 15 control patients (Figure 1) include 18 former smokers (Table 1)?
Figure 1 preserves the randomization schedule with analysis taking into account the exclusion of patients not meeting criteria (n=15 Control; n=21 Cell Therapy) (1). Table 1 reflects protocol defined analysis where all patients not receiving cell therapy were placed in the Control arm (n = 24), and all those receiving cell therapy placed in the Cell Therapy arm (n=21)(I). Both groups are described in the attached table.
6. Or were 9 control patients that reached Table 1 excluded before they reached Figure 1? If so, what was the reason and could the additional steps be explained in the Methods and shown in Figure 1 please?
The manuscript presents two complementary analyses as recommended in the peer-review process (1). Main figures maintain the randomization schedule with analysis reflecting the exclusion of patients not meeting criteria (n=15 Control; n=21 Cell Therapy). Online Figure 13 reflects protocol defined analysis where all patients not receiving cell therapy were placed in the Control arm (n = 24), and all those receiving cell therapy placed in the Cell Therapy arm (n=21) with patient demographics shown in Table 1. We here include an updated demographics Table with baseline characteristics pertinent to both analysis. Neither the 15/21 nor the 24/21 baseline comparisons reveal significant differences in populations, nor change the outcome of efficacy signals. The revised Table corrects for original typographical errors, which had no material impact on trial findings, interpretations or conclusions.
7. If there were 47 or 48 patients actually randomized, please could a revised Table 1 be presented showing their baseline characteristics, according to their randomization group?
Informed consent was requested from a total of 48 patients. As 1 patient refused participation, 47 patients were randomized.
8. Randomization ratio. Currently, the authors’ company website (Figure 1 below) tells us randomization was 1:1. There is nothing in the paper, the clinicaltrials.org (5)entry, nor the EudraCT (3) entry to suggest otherwise. The company webpage must be current because it hotlinks visitors directly to the JACC paper on ScienceDirect for the trial results. Can the authors confirm that these are indeed the same study, i.e. randomization was 1:1?
Randomization (2:1) was executed by a site-independent centralized process. The website of Cardio3BioSciences, the sponsor of the study, indicated at one stage a 1:1 ratio. This error has been corrected.
9. How did a 1:1 randomization ratio, produce a 15:32 distribution? The binomial probability of getting 32 or more randomized to stem cells from the 47 is <0.01. Was this an extreme chance effect? Or was randomization1:2?
As noted, randomization was 2:1.
10. How many cell therapy patients died? The article indicates (page 2334) “in the cell therapy group, one patient who underwent elective cardiac transplantation at 21 months post-randomization, postoperative sepsis developed, and the patient died.” There was no other death in the cell therapy group. How does this fit with an earlier report, that bone marrow harvest was performed in 31 of 31 patients, but only 30 out of 30 MSC isolations were performed, because “1 patient died”4 (see Figure 2). Can the authors please clarify this? Was that early death different from the 21- month death and if so, why was it not counted in the JACC paper?
As correctly stated in the manuscript (1), one patient died in the cell treated group of the randomized CCURE trial. This followed an elective cardiac transplantation, and post-operative sepsis. In a separate non-randomized pilot study, the “Swiss feasibility study” - distinct from the C-CURE trial - one patient died prior to being treated. The event is thus not reported in the manuscript, while this information may have been presented in earlier presentations to represent the totality of experience.
11. Ideally all events after randomization should be counted in the randomized arm. Could the authors report outcomes of patients according to intention to treat?
In the review process, it was suggested that intent to treat analysis for adverse events would create a delusive effect on any potential adverse events that could be attributed to cell therapy. Accordingly, events were reported based on patients who did and did not receive cell therapy. Regardless of mode ofanalysis, no statistical difference was noted for events between the cell therapy and control groups.
12. Study protocol. This trial is said to be randomized. In an earlier report (4) there was “one additional patient enrolled” in a non-randomized manner as part of the “Swiss feasibility study” (see Figure 3)? Exactly how many patients were randomized, and how many allocated nonrandomly? Is the C-Cure trial therefore not one trial but two, one of which is non-randomized, i.e., a cocktail?
The C-CURE trial is a single randomized trial. As indicated above, the “Swiss feasibility study” is separate.
13. The report from these authors of what appears to be the same study suggests that there was a partnership between Biotronik and Cardio3 Biosciences to implant into patients a special wireless Home monitoring ICD (5,6). Are these two the same trial? If so, why does this paper not mention this corporate involvement?
In the C-CURE trial, if patients were not already fitted with an implantable cardioverter-defibrillator, one was provided. In these cases, if criteria for country-specific insurance coverage were not met, a device was provided by the study sponsor, Cardio3 Biosciences, purchased at cost from a single supplier (Biotronik).
14. and 15. Results of safety evaluation. A safety analysis reported 27 adverse cardiac events after cell treatment (Table 2) including 15 supraventricular arrhythmias. How does this fit with an earlier report of 42 arrhythmic episodes for the cell therapy group “(2.0/patients)” (7)? Another earlier report of arrhythmias of the C-Cure Trial (8) stated that there had been 43 arrhythmic episodes in the recipients? Can cumulative events decrease with time? In the control arm, how do the 63 arrhythmic events in 18 controls reported earlier (7)fit the smaller number of arrhythmic events now reported in Table 2? How do 5 ventricular fibrillation episodes requiring ICD shock in the control group reported in 2011 (7) fit with just 2 in the control group, reported now? How do the number of 58 episodes of arrhythmic events in 18 control patients (“2.4/patient”) reported earlier (8), fit the number of arrhythmicevents reported in Table 2?
Initial communications presented data in various forms, e.g., the number of arrhythmic events versus the number of patients with arrhythmia. The manuscript reports those adjudicated by the DSMB (1).This evaluation identified and eliminated false positives (e.g., events recorded by the ICD as supraventricular tachycardia that were actually sinus tachycardia from physical activity), and grouped certain events. For example, recurrent non-sustained VTs over a short period of time or episodes sustained VTs that initially converted following anti-tachycardia pacing but then recurred. The beginning of a dysrhythmic event through its conclusion by cardioversion was considered one episode. Furthermore, the number of episodes versus the number of patients suffering from arrhythmia was distinguished in the manuscript (1). Table 2 of the manuscript also includes a listing of “new events” whereby events that pre-existed the index procedure are not deemed mutually independent (1). This conforms to regulatory agencies guidance on the causality assessment needed to determine that an adverse event may be caused by a drug/biological product (3). Indeed, the C-CURE study shows that cell therapy did not appear to be associated with an increase in new arrhythmic events (1).
16. The Tables show lower numbers of stem cell recipients suffering dyspnea and chest pain, than the authors previously reported. Can the authors please explain why some previously reported events are now not counted (9)?
The referenced slide 28 in a previous oral preliminary presentation indicates 1 patient suffering from dyspnea and chest pain. Table 2 in the manuscript also reports one case of dyspnea (I). This is not to be confused with dyspnea associated with cough due to upper respiratory infection.
17. Fractional patients in Table 1: The percentages in many cases contradict the counts, and sometimes suggest fractional patients. In other cases, the counts of patients in a subgroup of risk factors in the control group seem to
exceed that of the randomized control group. Moreover in several instances the stated percentages do not fit thestated number of patients. The inconsistencies are marked in Figure 2 below. Could they be corrected and explained please?
An updated Table is enclosed which provides an additional breakdown of control patients. The Table also corrects for two elements: one related to statistical misprogramming of the SAS data set; and a typographical mistake whereby patients under Statins should be 23 (96%) instead of 13 (96%).
18. Can the authors provide a clarification for the uncertainties in Table 2? The inconsistencies are marked below (see Figure 5). Could they be corrected and explained please?
A typographical error was noted in Table 2 where the number of patients in the treatment group (Post Cell Rx) of 6 is given a percentage of 20 which should read 28. There are also three rounding errors with 2 of 24 patients (8.33%) rounded to 8.4%, 2 of 24 patients (12.5%) given as (12.6%) and 2 of 21 patients (9.52%) rounded to 9.6%. As can be seen, those rounding errors are immaterial.
19. In Table 1 the p-values are calculated on the basis of there being 24 randomized controls. Yet Figure 1 shows that only 15 were randomized to the control arm. So the p values are meaningless for a reader interested in the randomized controlled trial. Please could the authors provide the comparisons for the actual randomized patients?
As indicated, a Table is included encompassing both analysis (15/21 and 24/21) with appropriate p values.
20. The same correction, an updated set of p values based on the actual randomized patients, is required for Table 2.
Presented p values in Table 2 are accurate.
21. Can the authors explain why Figure 4a only has 15 controls, if there were 22 in Table 1?
We note that Table 1 mentions 24 controls (not 22 as claimed in this question). The explanation regarding the control group has been provided in response to questions 4, 5, 6, 8, 9, 11 and 19. The manuscript presents two complementary analyses (1). Main figures, including Figure 4a, maintain the randomization schedule with analysis taking into account the exclusion of patients not meeting criteria. Online Figure 13 reflects analysis where all patients not receiving cell therapy were placed in the Control10 arm, and all those receiving cell therapy placed in the Cell Therapy arm with patient demographics shown in Table 1.
22. Figure 4B shows the mean EF of cell recipients increases from 27.5% to 34.5%. The increments in EF in Figure 4A for cell therapy must therefore average 7.0%. However the graph shows the average increment is (2+1.5+8+4+4.25+5+4.25+8+2+8+9+7.5+3+10+7+6+8+2+7+1.5+6) / 21 = 5.4%. Incidentally our figure reading resembles the value in a previous report10 of these data, which indicated a 5.2% increase of LVEF in recipients. Which is correct, the current graph (and previous report) or alternatively the current text and its abstract?
Median ejection fraction reported in the manuscript and described in the legend to Figure 4 indicates a statistically significant improvement following cell therapy as illustrated in Figure 4B; Figure 4A provides per patient data to highlight the range of effect following cell therapy with no patient showing signs of deterioration in this parameter (1). The values reported in the manuscript accurately reflect data (1);values mentioned in the comment are a departure from the actual values. Importantly, signals of benefit on cardiac parameters without clinical deterioration in cell treated patients corroborated the safety profile of the stem cell-based intervention.
23. The methods state that “the defining inclusion criterion was chronic heart failure of ischemic origin with impaired left ventricular ejection fraction (LVEF) (15% to 40%)”. Figure 1 shows that the screening process removed patients with EF>45%. What happened to patients with EF 40-45%?
The inclusion criteria defining the C-CURE study included a left ventricular ejection fraction (LVEF) >15% and ≤40%. Accordingly, 6 screened patients were excluded from further consideration as their ejection fraction actually exceeded 45%.
24. and 25. Please explain Figure 5c. The percentage of patients with changes in MNQOL score ≥10 points is stated to be zero for the control patients. How did the authors calculate the standard deviation or confidence interval? Following on from this, by exactly what calculation steps have standard deviations or confidence intervals been obtained for the classification groupings for the 10 variables: Survival, Hospitalisation, NYHA, MNQL, Walk, VO2 max, Mass index, EF, ESV, EDV?
Figure 5C depicts the percentage of patients showing improvement or deterioration in ten clinical parameters. For each of the followed clinical parameter and using established thresholds, a scoring system was assigned to track patients showing signs of improvement, no change or deterioration, respectively. Thresholds were set as change of 10 ml in left ventricular end diastolic/end-systolic volume, 5 g/m2 in left ventricular mass index, 5% in ejection fraction, 20 m in 6-min walking distance, 2 ml/kg/min VO2max, 10 points in Minnesota Living With Heart Failure Questionnaire, 1 functional class in New York Heart Association classification, and presence/absence of hospitalization and mortality (1).Online Table 3 provides a break-down of patient groups for each parameter (1).This algorithm integrates values obtained for all patients and provides a fixed range of distribution (Figure 5C) (1).Positive percent values indicate improvement, while a negative value indicates patients that showed deterioration. Of note, patients with lower scores in the Minnesota Living With Heart Failure Questionnaire (i.e., Δ≥-10) are considered improved, and as such were assigned a positive score. Moreover, when patients had a NYHA class change of two, a score of two was assigned. Safety endpoints evaluated in this calculation took into account a 24-month observation period; efficacy endpoints were tracked at 6 months.
26. Why do all error bars in Figure 5c appear to have the same size?
Although bar lengths in Figure 5C may appear of same size, they are not.
27. How can there be negative percentages of control patients with changes in walk distance? And of controls and recipients with changes in peak oxygen uptake?
Negative percent values indicate patients that showed deterioration.
28. and 29. Figure 5c indicates that at 6 months survival in controls is 90%. Yet both Figure 1 and the article text (page 2334) state clearly that the two deaths amongst the controls were after 6 months, so should this not be 100%? Can the authors please clarify which is correct? Could they similarly clarify the discrepancy in survival and mortality values in Appendix Online Table 3.
In Figure 5C and online material, mortality and hospitalization data are at 2 years as described in the Methods (1). This should have been made clear in respective legends to be consistent with main text (page 2334). Values are accurate in the manuscript figures and main text.