Kisspeptin-54 Second-Dose IVF Phase 2 Trial: NCT01667406

This 62-person phase 2 trial did not compare kisspeptin-54 with no kisspeptin. Every randomized participant received a first kisspeptin-54 exposure intended to trigger oocyte maturation; randomization determined whether a second exposure or saline followed 10 hours later. The second-exposure group more often crossed the paper’s stated primary threshold of mature-oocyte yield at least 60%, but the result rests on small counts and a debated analysis history. The study was not powered to establish differences in pregnancy, live birth, ovarian hyperstimulation syndrome, or uncommon harm, and it cannot validate another kisspeptin form or product.[1][2][3][4]

Study at a glance

The protocol contains exact exposure amounts and a wider IVF regimen. Those historical details identify the experiment only; they are not fertility-treatment, preparation, timing, or administration guidance.

The randomized question was the second exposure

All 62 randomized women received kisspeptin-54 first. Thirty-one were assigned to another kisspeptin-54 exposure 10 hours later, while 31 received saline then. Describing the arms as “kisspeptin versus placebo” would be materially wrong because it erases the common first exposure. The causal contrast, if valid, is the incremental effect of adding the second exposure within this protocol.[1]

The study enrolled women considered at increased risk of ovarian hyperstimulation syndrome, defined by anti-Müllerian hormone at least 40 pmol/L or antral-follicle count at least 23. It took place at Hammersmith Hospital, Imperial College London. Printed allocation lists were prepared by a statistician; participants, embryologists, and IVF clinicians were masked to the second exposure. This narrow, single-center population limits generalization to people with different ovarian reserve, ages, infertility diagnoses, or IVF protocols.

Participant flow and registry scope

Seventy-six women were screened. Ten were ineligible, two withdrew before the protocol, and two encountered conditions outside the protocol before randomization. The remaining 62 were randomized, 31 per group, and all completed the primary study procedures. That clean randomized flow does not make the trial large enough to resolve uncommon safety outcomes.[1]

ClinicalTrials.gov reports 175 participants because NCT01667406 is an umbrella kisspeptin IVF protocol containing multiple cohorts and exposure groups. The 175-person registration population must not be presented as the sample for this second-dose comparison. Conversely, adverse-event totals from every registry cohort cannot automatically be attributed to these 62 women.[2]

Primary oocyte-yield result and an internal count discrepancy

The primary outcome was a binary threshold: whether mature-oocyte yield reached at least 60%. Yield was the number of mature metaphase-II oocytes divided by the number of follicles at least 14 mm on the morning of the first study exposure. This endpoint is not simply the total number of eggs retrieved, and crossing 60% does not guarantee fertilization, pregnancy, or live birth.[1]

Table II reports 14/31 women, or 45%, crossing the threshold after the single kisspeptin-plus-saline sequence and 22/31, or 71%, after the two-kisspeptin sequence. The abstract prints 21/31 alongside 71%; 21/31 is not 71%, while 22/31 is. The table’s 22 is arithmetically consistent with the percentage and underlies the displayed comparison, but the article does not explain the one-participant textual discrepancy. It should remain documented rather than silently corrected.

The reported absolute difference was 25.8 percentage points (95% confidence interval 2.1 to 49.5), with odds ratio 2.97 (95% CI 1.04 to 8.48) and p=.042. The lower confidence limits sit close to no difference. In a sample of 62, moving one or two classifications could materially affect the inference.

Published correspondence about endpoint timing and analysis

After publication, Hill and colleagues questioned whether the 60% dichotomous endpoint had been publicly specified before analysis. They noted that the older registry wording was broader and calculated a two-sided Fisher exact p value of .071 from the small table, with an odds-ratio interval that included no effect. They argued that the timing and choice of analysis needed greater transparency.[3]

The investigators replied that the endpoint and logistic-regression analysis were specified in a statistical analysis plan before unmasking, that the power calculation used that endpoint, and that an independent statistician performed the analysis. A protocol and statistical-analysis-plan file are now linked from the registry, but structured results were posted in 2021, after the article and correspondence.[2][4]

These publications support a careful conclusion, not an accusation or a dismissal. Logistic regression and Fisher’s exact test answer the small-table question differently, and the public chronology is less transparent than a prospectively time-stamped registry entry would have been. Readers should see the effect estimate, interval, small counts, analysis choice, and published exchange together.

Embryology and reproductive outcomes

Fertilization rates were similar, approximately 79% in each group. Implantation was 23.3% after the single-exposure sequence and 37.0% after the second-exposure sequence; the reported odds ratio was 2.1 with a wide 95% CI of 0.68 to 6.3. Clinical pregnancy occurred in 7/31 versus 12/31 women, an absolute difference of about 16 percentage points whose confidence interval crossed zero. Live birth occurred in 6/31 versus 12/31.[1]

Those pregnancy and live-birth counts may be clinically interesting, but they were secondary and imprecise. The study was designed around the oocyte-yield threshold, not powered to demonstrate a live-birth benefit. Embryo-transfer choices, cycle segmentation, and post-randomization reproductive steps also separate the trigger response from later outcomes.

Ovarian hyperstimulation and other adverse events

One woman in the single-exposure/saline group had moderate early ovarian hyperstimulation syndrome and was admitted for less than 24 hours for analgesia; the paper says no further medical intervention was required. One woman in the second-exposure group had mild late ovarian hyperstimulation syndrome. With one event per group, the trial cannot estimate a reliable between-group difference or establish prevention.[1][2]

The paper describes study exposures as generally well tolerated. One woman in the saline-second group had an unexplained miscarriage at 19 weeks; ClinicalTrials.gov represents a serious event in that arm and none in the second-kisspeptin arm for the relevant randomized groups. The broader registry reports events across the umbrella study over several years, so cohort attribution must be checked rather than pooled casually. No 62-person study can exclude uncommon maternal, reproductive, or offspring harms.

Sponsor, funding, and disclosures

Imperial College London sponsored the registered study. The work reports support from the Medical Research Council, Biotechnology and Biological Sciences Research Council, National Institute for Health Research, Wellcome-related Imperial infrastructure, and investigator awards. The paper says the authors designed, conducted, analyzed, and reported the study and declared no competing interests.[1][2]

Public registry materials were updated after the paper, so current files help reconstruct the protocol but do not erase the importance of when endpoint definitions became publicly visible.

Statistical significance, effect, uncertainty, and importance

• Statistical significance: the planned logistic model produced p=.042, while the correspondents’ unadjusted Fisher exact calculation produced p=.071.
• Effect size: the table shows 14/31 versus 22/31 crossing the 60% yield threshold, with an estimated absolute difference of 25.8 percentage points.
• Uncertainty: the interval ranged from 2.1 to 49.5 points, the abstract/table count differs by one, and the cohort came from one center.
• Clinical importance: the endpoint was oocyte yield; pregnancy, live birth, ovarian-hyperstimulation prevention, and long-term safety were not established.

Important limitations

• Both randomized groups received kisspeptin-54, so the trial cannot estimate its effect versus no kisspeptin.
• Sixty-two participants provide limited precision and very little power for uncommon adverse events.
• The 60% threshold and analysis chronology prompted published methodological correspondence.
• The abstract and primary table differ between 21 and 22 second-exposure responders.
• Pregnancy and live-birth outcomes were secondary and underpowered.
• One specialist center and high-risk selection constrain external validity.
• Registry enrollment of 175 belongs to the umbrella protocol, not this comparison alone.

What this study does not prove

NCT01667406 does not prove that kisspeptin-54 is superior to no kisspeptin, guarantees mature oocytes, improves live-birth probability, prevents ovarian hyperstimulation syndrome, or is broadly safe across IVF populations. It supports a limited second-exposure signal on one threshold endpoint under one protocol.[1][2][3][4]

It also does not establish equivalence to kisspeptin-10, another peptide fragment, another formulation, a compounded preparation, or a catalog material. Sequence name alone cannot transfer identity, purity, stability, or clinical findings.

Frequently asked questions

Did the placebo group receive no kisspeptin-54?

No. Every randomized participant received the first kisspeptin-54 exposure; saline replaced only the second scheduled exposure.[1]

How many women were in the randomized comparison?

Sixty-two women were randomized, 31 per group, and completed the primary procedures. The registry’s 175 participants cover the wider umbrella protocol.[1][2]

What did the primary endpoint measure?

It measured whether mature metaphase-II oocytes represented at least 60% of follicles at least 14 mm before the first exposure.[1]

Were there 21 or 22 primary responders in the second-exposure group?

Table II reports 22/31 and 71%; the abstract prints 21/31 with 71%. The table count matches the percentage, but the paper does not explain the discrepancy.[1]

Why did published statisticians question the result?

They questioned public endpoint timing and showed that a Fisher exact test did not cross .05. The authors replied that a prespecified plan called for logistic regression and the 60% threshold.[3][4]

Did the study prove more live births or less ovarian hyperstimulation?

No. Live birth and ovarian-hyperstimulation outcomes involved small secondary counts, and the trial was not powered for those conclusions.[1][2]

Does this trial validate kisspeptin-10 or a catalog peptide?

No. It studied native kisspeptin-54 within a defined clinical protocol and cannot authenticate or establish comparability to another sequence, formulation, supplier, or lot.

Primary and official sources

1. Primary full article, PMCID PMC5850304
2. ClinicalTrials.gov: NCT01667406
3. Published correspondence, PMID 29194480
4. Investigators’ reply, PMID 29194495
5. PubMed: PMID 28854728
6. Publisher version of record, DOI 10.1093/humrep/dex253