[Edaily Reporter KIM JI-WAN ] Kim Seung-won, the nominee for Minister of Justice, addressed the controversy surrounding the request for expedited processing of clinical trials for GenenCell’s COVID-19 treatment “ES16001,” explaining, “I reviewed data showing that the treatment was effective, based on overseas clinical trials that had progressed to Phase 2.” As a result, the scientific basis for the “therapeutic efficacy” claimed by GenenCell at the time is now coming under renewed scrutiny.

On October 12, 2021, while serving as a member of the National Assembly, Nominee Kim contacted Kim Kang-rip, then Commissioner of the Ministry of Food and Drug Safety (MFDS), asking him to “keep a close eye on” Genencell’s Investigational New Drug (IND) application for a COVID-19 treatment. The Ministry approved the Phase 2 and 3 clinical trials 14 days later, on the 26th of the same month. Candidate Kim’s camp maintains that this was merely a request to address a public interest concern and that no preferential treatment was granted in the approval process.

However, an analysis of the paper detailing the results of the Phase 2 clinical trial in India—which served as the basis for the determination that the treatment was “effective” at the time—raises further doubts.

Accordingly, over two days on the 7th and 8th, E-Daily verified the validity, safety, and appropriateness of the statistical analysis in the paper detailing the clinical results of Genencell’s COVID-19 treatment through multiple experts, including Mr. A, the current director of a biotech research institute with experience in developing COVID-19 treatments.

Genencell’s COVID-19 Paper… “Not a Formal Paper” The paper

containing the Phase 2 clinical trial results for Genencell’s COVID-19 treatment “ES16001” is labeled “EBioMedicine Article” on its first page. However, the document itself remains a preprint that has not yet undergone peer review.

At the bottom of every page of the 39-page paper, the statement “This preprint research paper has not been peer reviewed” is repeated.

A preprint is a paper in which a researcher publishes their research findings before they undergo the formal review process of an academic journal. While it offers the advantage of allowing research results to be shared quickly, it differs from a formally published paper in that it has not undergone peer review—a process in which experts in the field verify the research design, statistical analysis, and interpretation of results.

Mr. A, the current director of a biotech research institute with experience in developing COVID-19 treatments, pointed out on the condition of anonymity, “E-Biomedicine is an influential journal affiliated with The Lancet, with an Impact Factor (IF) of 11,” but added, “The document currently available does not appear to be a paper that has been accepted for publication, but rather a manuscript at the submission stage.”

He went on to emphasize, “Since it has not undergone peer review, it is correct to view this as merely the research team’s claim for now.”

The Lancet is one of the world’s most prestigious general medical journals, renowned for its rigorous peer review and high academic influence.

The paper was published in January 2022 on SSRN, a preprint platform. As far as can be publicly verified, there is no evidence that these Phase 2 clinical trial results subsequently underwent peer review and were published as a separate final paper.

The fact that it is a preprint does not necessarily mean the research findings are incorrect. However, the fact that a study presenting strong efficacy data—such as “a recovery rate of 95% versus 68% on Day 6, P=0.00021”—has not been confirmed as a peer-reviewed final paper even now, four years and eight months later, is a significant limitation when evaluating the reliability of the results.

Mr. A stated, “If this were officially published data, we could evaluate the research methods, statistics, and interpretation in greater detail, but for now, we must take into account that this data has not yet been verified by external experts,” expressing doubts about the reliability of the entire paper.

India: Cost Advantages, but Questions About Early Clinical Trial Reliability

The fact that Genencell’s Phase 2

clinical trial

was conducted in India is another point to consider when interpreting the results.

Mr. A noted, “India offers the advantage of being able to recruit a large number of patients at a relatively low cost, which is beneficial for large-scale Phase 3 trials,” but added, “Since Phase 1 and 2 trials must be conducted precisely in a controlled environment, the expertise of the trial sites and their data management capabilities are crucial. In this regard, we must closely examine the differences in capabilities among the participating institutions.”

In particular, this Genencell clinical trial involved only 60 patients in total. Given that the analysis was further subdivided into mild and moderate subgroups, even minor differences—such as patient classification criteria, consistency in standard-of-care treatment, symptom assessment methods, and adherence to assessment timelines—could have a relatively significant impact on the statistical results.

The actual GenenCell clinical trial was conducted on a total of 60 patients. All 60 patients were randomized into the ES16001 plus standard treatment group (30 patients) and the placebo plus standard treatment group (30 patients); three patients dropped out from each group, resulting in 27 patients completing the trial in each group, for a total of 54 patients. The reasons for withdrawal were the same in both groups: 2 cases of withdrawal of consent and 1 case of failure to follow up.

The issue is that these 54 patients were further divided into mild and moderate subgroups for subgroup analysis. Although the paper reported statistical significance in the mild subgroup, it did not clearly specify how many patients were included in each of the mild and moderate subgroups.

Mr. B, who is in charge of statistical analysis for clinical trials at a multinational healthcare company, pointed out, “This was a clinical trial that divided patients into subgroups even though the total number of patients was small to begin with,” adding, “Small differences—such as patient classification criteria, symptom assessment methods, variations in standard treatment, or management of assessment timepoints—can have a relatively large impact on the results.”

Therapeutic efficacy... really?

The Genencell research team conducted a randomized, double-blind, placebo-controlled Phase 2 clinical trial at SPARSH Hospital in India involving 60 patients with mild to moderate COVID-19.

Thirty patients received ES16001 in addition to existing standard treatment, while the remaining 30 received a placebo alongside standard treatment. According to the paper, by day 6 of treatment, 95% of patients in the ES16001 group had recovered from their symptoms, compared to 68% in the placebo group. The researchers explained that a Kaplan-Meier analysis comparing the time to recovery showed a P-value of 0.00021, indicating a statistically significant difference.

The issue is that the results differ when patients are grouped by symptom severity.

Among patients with mild symptoms, the ES16001 group recovered significantly faster than the placebo group, with a P-value of 0.00064. In contrast, no significant difference was observed among patients with moderate symptoms. The P-value presented in the paper’s graph is 0.091. Patients with severe symptoms were not included in the trial from the outset.

The research team also titled the results section of the paper “ES16001 resulted in a high recovery rate in COVID-19 patients with mild symptoms.”

Mr. A stated, “In reality, most patients with mild COVID-19 recover over time,” adding, “While an effect that prevents deterioration or significantly shortens the recovery period in high-risk groups may have therapeutic value, further verification is needed before strongly asserting the drug’s efficacy based solely on the results of this small subgroup of mild cases.”

Mr. A also drew attention to the paper’s explanation of the mechanism of action. He said, “The researchers explain that the effect of inhibiting viral replication observed in cell experiments was reflected in clinical results, but inhibiting viral replication can be viewed as an effect primarily seen in the relatively early stages of infection,” adding, “Based on the paper’s results, there is room to interpret that the effect was not pronounced in the moderate stage, when inflammation and pneumonia have fully developed.”

Abstract States “Significant Reduction in Inflammation”… But the Discussion Tells a Different Story

A closer look at the biomarker section only serves to amplify doubts about the therapeutic efficacy of ES16001.

The paper’s abstract states that ESE “significantly reduced levels of inflammatory mediators.”

However, the “Discussion” section presents a completely different narrative.

The researchers wrote that TNF-α “tended to decrease” compared to the placebo group. Regarding PGE2 and IL-6, they explicitly stated, “ES16001 treatment did not significantly alter…compared to the placebo group”—meaning there were no significant changes compared to the placebo.

Looking at the actual results, PGE2, TNF-α, and IL-6 levels dropped significantly over time not only in the ES16001 treatment group but also in the placebo group. Therefore, a distinction must be made between “within-group significance”—a decrease compared to pre-treatment levels—and “between-group significance”—superior efficacy compared to the placebo.

Researcher C, who has conducted new drug development and preclinical research at a domestic pharmaceutical company, analyzed the data, stating, “Cytokine expression may be inconsistent during the early stages of infection,” but added, “Considering the virus replication inhibition mechanism claimed in the paper, it appears that the efficacy was relatively lower at the stage where the condition progressed to moderate severity and the inflammatory response became fully established.”

Enrollment occurred on Days 1, 5, and 10… So why is Day 6 the paper’s “key figure”

? There is also a point worth noting in the statistical analysis.

The primary endpoints for the Phase 2 ES16001 clinical trial, registered with the Indian clinical trial registry CTRI, include recovery from clinical symptoms, the DASS, the Hamilton Anxiety Scale, the WHO-5, RT-PCR, and PGE2, TNF-α, and IL-6. The evaluation timepoints were registered as Day 1, Day 5, and Day 10, with a target patient enrollment of 60.

However, the efficacy figure most emphasized in the paper is “a recovery rate of 95% versus 68% on Day 6, P=0.00021.”

Analyzing the data from Day 6 is not inherently problematic. It is a valid approach if the patients’ conditions were monitored daily and the statistical analysis plan (SAP) developed prior to the start of the clinical trial included an analysis of “time-to-recovery.”

Mr. B noted, “We cannot definitively conclude that it is wrong,” but added, “There are aspects where the criteria for observation dates and evaluation dates are unclear.”

Ultimately, the key issue is whether the analysis of the recovery rate on Day 6 was predetermined before the trial began. If this was a post-hoc analysis conducted after the trial ended, its evidential weight may be lower than that of a pre-specified analysis.

Main Text States “No Difference in Safety”… Table 4 Shows P<0.05

Discrepancies requiring explanation were also found between the main text and the tables regarding safety data.

In the main text, the research team described the results of their analysis of CBC, urinalysis, and blood chemistry tests, stating, “There were no statistically significant differences within or between groups for any safety指标,” and concluded that ES16001 is safe.

However, Table 4 actually lists multiple instances of P < 0.05.

For example, when comparing baseline values to end-of-treatment (EOT) values, the intergroup comparison for RBC between Arm B and the other group is marked as P < 0.05. For AST and ALT, P < 0.05 was observed in Arm A, while for creatinine, P < 0.05 was observed in both arms. When comparing baseline values to Day 5, P < 0.05 was also observed for HDL, LDL, creatinine, uric acid, and urine pH.

In particular, while statistically significant changes in the liver function markers AST and ALT were observed only in the treatment group (Arm A) administered ES16001, no significant changes were observed in the placebo-treated control group (Arm B). However, this alone is not sufficient to conclude that there was a problem with patient randomization.

However, to determine whether the two groups were balanced in terms of AST and ALT levels from the start of the trial, specific data—such as the mean and variance of baseline values—are required. Since the current table presents only P-values without actual measured values, it is difficult to ascertain whether the levels increased or decreased, or to determine the magnitude of the changes.

This does not necessarily mean that ES16001 is unsafe. Repeated comparisons across various test items can yield random statistical differences, and statistical significance is a different concept from clinically meaningful adverse events. In fact, the research team stated that there were no deaths, serious adverse events (SAEs), or adverse events leading to trial discontinuation.

The issue lies in the paper’s explanation. While the table shows statistically significant changes, the main text categorically states that “there were no significant differences” and fails to adequately explain why these differences were deemed not to pose a safety concern.

After reviewing Table 4, Mr. A pointed out, “It is quite unusual to see a table in a clinical paper that lists only P-values without actual numerical values,” adding, “Especially when safety-critical markers like AST and ALT show statistically significant changes in only one group, it is necessary to present the actual values and the distribution of baseline levels to provide an explanation.”

AST and ALT are key blood test markers used to assess liver cell damage and are crucial in new drug clinical trials for identifying potential drug-induced hepatotoxicity.

He assessed, “Ultimately, what a clinical paper must clearly demonstrate through data is efficacy and safety,” adding, “This paper includes early-stage research—ranging from computer-based analysis (in silico) and in vitro experiments to animal studies—in great detail, yet the aspects of efficacy and safety, which are most important in clinical trials, remain unclear.”