Reproductive immunology
Home · Research areas

Recurrent early pregnancy loss

Recurrent early pregnancy loss is spoken of from three early pregnancy losses onwards. When the validated work-up comes back normal, the question becomes an immunological one. Arsène Mekinian is principal investigator of the FALCO national registry, which follows these situations prospectively, and last author of the 2024 publication that drew the first result from it, which is negative.

Recurrent early pregnancy loss, that is repeated miscarriage before the end of the first trimester, is a frequent and complex clinical situation, whose assessment rests on a multidisciplinary approach bringing together obstetrics and gynaecology, reproductive medicine, genetics and, in certain situations, internal medicine and clinical immunology.

A substantial proportion of patients remain without an identified cause after the recommended work-up. One of the current challenges of research is to determine whether, in some of them, immunological or inflammatory mechanisms take part in the repetition of pregnancy loss and can constitute therapeutic targets.

What establishes his expertise in this field

What he coordinates or directs.

  • Principal investigator of the FALCO national registry, Cohort of Patients Presenting Unexplained Recurrent Miscarriages and Identification of Early Miscarriage Recidivism Factors, NCT05557201, sponsored by AP-HP, according to the public registry entry.
  • Director since 2024 of the diplôme d'université d'immunologie de la reproduction (university diploma in reproductive immunology) of Sorbonne Université, according to the page for the diploma published by Sorbonne Université Formation Continue, updated on 28 July 2026.
  • He states that he provides the coordination of the réunion de concertation pluridisciplinaire nationale « Infertilité » ("national multidisciplinary team meeting on infertility") of the FAI²R network. The network lists this meeting among its national meetings and names no coordinator.

His position in the publications cited here. Of the 12 references cited at the end of this page, Arsène Mekinian is first author of 4, last author of 6, and corresponding author of 5. Two further references are cited in the section on what remains to be shown without his being a signatory: a randomised trial published by a Chinese team and a comment letter addressed to one of his publications, both of them needed to set out the state of the discussion.

In biomedical research, the first author is generally the person who carried out the work day to day, and the last author the one who designed and supervised the study. The literature describes this as an informal practice rather than a rule: international recommendations leave the order of authors to the authors themselves, and reading seniority from position becomes unreliable as the list of authors grows. [How to read author position](/en/method/author-position/)

Definition and work-up

A missed pregnancy is a pregnancy that stops of its own accord. The definitions of recurrent miscarriage differ between learned societies and between research protocols, in particular on the number of pregnancy losses required before investigations are started. The French national registry and the French work on the subject take at least three losses occurring before 14 weeks of gestation; some learned societies open the work-up from two losses onwards, and this divergence of definition is one of the reasons why studies in the field compare poorly with each other.

The work-up aims first to look for the main recognised causes or associated factors: abnormalities of the uterine cavity, a chromosomal abnormality in either parent, endocrine or thyroid causes and antiphospholipid syndrome, taking account of the obstetric history and of the clinical context.

When this assessment does not identify a cause, the miscarriages are considered unexplained in the current state of knowledge and of validated investigations. In the first publication from the FALCO registry, the situation remained unexplained for 76 of the 100 pregnancies analysed after the work-up recommended by ESHRE. [Ref. 1]

This situation does not mean that there is no underlying mechanism. It defines precisely the field in which new pathophysiological hypotheses have to be assessed. It is at this point that internal medicine comes in alongside obstetrics and gynaecology, with a precise hypothesis: in some of these women, the interruption of pregnancy would be due to a defect in the mother's immune tolerance towards the embryo, which is genetically half-foreign.

What place for immunology?

Implantation and placental development require close regulation of the interactions between the maternal immune system, the endometrium and the trophoblast.

Several mechanisms have therefore been studied in recurrent miscarriage: autoimmunity, NK cells, regulatory T cells, cytokines, HLA/KIR interactions, endometrial inflammation and inflammatory abnormalities of the placenta.

The mechanisms documented by his own publications are autoimmunity, circulating and uterine NK cells, regulatory T cells, cytokines, the immune profile of the endometrium and inflammatory abnormalities of the placenta.

However, demonstrating an immunological abnormality does not necessarily mean that it is responsible for the pregnancy losses. The work carried out in this area shows how difficult it is to identify an immunological biomarker reproducible and predictive enough to be used on its own to guide management. On this point, his clearest piece of work is a negative result which he signs as last author and corresponding author.

Published in 2019 in Archivum Immunologiae et Therapiae Experimentalis, this work combines a case-control study and a meta-analysis of the literature. One hundred and fifteen women were recruited consecutively from December 2015 to October 2017 in three university hospitals: 54 with recurrent miscarriage, 41 with repeated implantation failure, and 20 controls with at least two term deliveries. The proportions of circulating CD3-CD56+ NK cells and of CD8+CD57+ T-cell large granular lymphocytes measured before conception differed neither between cases and controls, nor between the women who subsequently miscarried and those who had a live birth. The meta-analysis of the literature, for its part, did find higher proportions of NK cells in recurrent miscarriage, mean difference 3.47 with an interval of 2.94 to 4.00, and in implantation failure, mean difference 1.64 with an interval of 0.82 to 2.45, both significant at p below 0.001, but with high heterogeneity between studies. The article concludes that the proportion of these cells in blood before conception does not reflect the risk of miscarriage or of implantation failure and that it should not be used as an indicator for management. He is 16th and last of the 16 named authors. [Ref. 3]

The scientific question today is therefore to determine which mechanisms are genuinely involved, in which patients and with what therapeutic consequence.

Lines of research

The work of Arsène Mekinian and of the associated teams is organised around several lines.

Characterising patients. Describing the clinical, obstetric, immunological and laboratory phenotypes associated with unexplained forms. This line is carried by the FALCO registry, described in the next section, and, before it, by a multicentre cohort published in 2021 in the American Journal of Reproductive Immunology: 101 women with at least three early miscarriages without aetiology, recruited in three university hospitals, and an analysis covering 652 pregnancies. Arsène Mekinian is 21st and last named author of it. The article was the subject of a comment letter published in the same journal in 2022. [Ref. 2]

Searching for biomarkers. Assessing the diagnostic or prognostic value of the various immunological parameters proposed. In the 2021 cohort, none of the immune markers measured differed between the women according to pregnancy outcome: neither NK cells, nor lymphocytes, nor gammaglobulins, nor blood cytokines, nor the activation status of uterine NK cells. The only difference observed concerned antinuclear antibodies, more often positive in the women who had a live birth, 12 cases that is 13% against 36 cases that is 7%, p = 0.03. [Ref. 2] A neighbouring piece of work, in which he is 9th of the 11 named authors, addressed the immune profile of the endometrium: a retrospective analysis of 104 patients included from January 2012 to December 2019, in which a dysregulated endometrial immune profile, then treated accordingly, was associated with a live birth rate of 55% against 45%, p = 0.01, with an odds ratio of 3.4, interval 1.27 to 9.84, independently of age and of anti-Mullerian hormone level. The declaration of interests of this article states that the last author created the company that markets this test and holds the patent covering it, and that two co-authors were employees of that company. [Ref. 4]

Autoimmunity and inflammation. Studying the links between autoimmunity, maternal inflammation, placental abnormalities and recurrence of pregnancy loss. This line covers the analysis of autoantibodies in the FALCO registry, described below, and the 2026 report of a pregnancy carried to term in a woman with massive perivillous fibrinoid deposition. [Ref. 1, 11]

Therapeutic assessment. Determining whether certain immunomodulatory strategies can improve obstetric outcome in precisely defined subgroups. This is the subject of the next two sections.

FALCO: a prospective cohort

FALCO, NCT05557201 is a prospective cohort sponsored by AP-HP devoted to recurrent miscarriage and repeated implantation failure. It aims to characterise the patients, the abnormalities found during the work-up, the treatment strategies used and the outcome of subsequent pregnancies.

The registry classes it as an observational cohort, under the full name Facteurs de récidive précoce des fausses couches, sponsored by Assistance Publique-Hôpitaux de Paris, and declares Arsène Mekinian principal investigator.

The registry entry shows an actual start on 24 October 2022, a planned enrolment of 700 participants, an estimated primary completion in October 2027 and an estimated study completion in October 2029, a recruiting status, and a last update posted on 19 December 2025. The inclusion criteria are an age of 18 to 50 years and at least three early pregnancy losses before 14 weeks of gestation, or at least three implantation failures. The declared primary objective is to describe the clinical and laboratory characteristics of these patients, the investigations carried out and the care received; the secondary objectives concern the recurrence rate according to the number of previous losses, according to the abnormalities found in the work-up and according to the treatments used, and the comparison of the efficacy of immunomodulatory treatments with each other. The 2024 publication states for its part that the registry has been established since 2017 and that it recruits in several university hospitals, naming Saint-Antoine Hospital, Tenon Hospital and Trousseau Hospital. [Ref. 1, entry NCT05557201]

The first published data illustrate the limitations of observational studies when it comes to attributing a benefit to an immunomodulatory treatment, and they reinforce the need for prospective controlled assessments. The first published analysis from this registry appeared in September 2024 in Human Reproduction. It covers 100 pregnancies occurring in 74 women exposed to hydroxychloroquine, marketed as Plaquenil, before conception or at the very beginning of pregnancy. The mean age of the women was 34.2 years, the median number of previous miscarriages was five, with a maximum of twelve. Women meeting the classification criteria for antiphospholipid syndrome were not included in the registry. Hydroxychloroquine was prescribed at 400 mg per day, before conception in 71% of cases; 78% of the pregnancies also received prednisone, 56% low-dose aspirin and 41% a low molecular weight heparin.

The main result is as follows: 62 pregnancies out of 100, that is 62%, ended before twelve weeks of gestation, 38 went beyond that term, and 31 out of 99 documented pregnancies, that is 31.3%, resulted in a live birth. The only factor significantly predictive of a pregnancy going beyond twelve weeks, after adjustment for age and for duration of exposure before conception, was having had four or fewer previous miscarriages, with an adjusted odds ratio of 3.13, 95% confidence interval 1.31 to 7.83, p = 0.01. The presence of at least one positive autoantibody did not change the outcome: 18 of 51 pregnancies went beyond twelve weeks in women with an autoantibody and without ovarian insufficiency, against 18 of 45 in those without, p = 0.63. The authors conclude that exposure to hydroxychloroquine in early pregnancy in women with recurrent pregnancy loss does not appear to prevent further miscarriages. They themselves flag the absence of a control group as the first limitation of their work. Arsène Mekinian is 14th and last of the 14 named authors. [Ref. 1]

From translational research to therapeutic trials

Several immunomodulatory strategies have been studied in unexplained miscarriage, in particular corticosteroids, hydroxychloroquine, intravenous immunoglobulins, intralipid and various approaches targeting immune regulatory pathways. The available data remain heterogeneous and of varying levels of evidence. The associations observed in cohorts or small series do not, on their own, establish the efficacy of a treatment.

The 2021 cohort gives the measure of this. Two hundred and fifteen pregnancies, that is 33%, were treated there with aspirin or low molecular weight heparin, alone or combined with various lines of immunomodulatory treatment. The cumulative live birth rate was 43.0% for treated pregnancies against 34.8% for untreated pregnancies, p = 0.04, and pregnancies on corticosteroids were twice as likely to result in a live birth as untreated pregnancies, odds ratio 2.0, 95% confidence interval 1.1 to 3.7, p = 0.02. This result is an association observed in a cohort in which treatment was not randomised: it does not demonstrate that the treatment is its cause. [Ref. 2]

The rest of this body of work is explicitly of a lower level of evidence, and says so itself. His 2016 review in the American Journal of Reproductive Immunology, of which he is 1st of the nine named authors, gathered the data available on corticosteroids, progesterone, intralipid, TNF alpha antagonists, G-CSF, hydroxychloroquine and intravenous immunoglobulins, and found by meta-analysis only a modest benefit of progesterone on live birth, odds ratio 1.38, interval 1.07 to 1.77, with significant heterogeneity, p = 0.01 and I² = 78%. [Ref. 5] His 2017 French review in La Revue de médecine interne, of which he is also first author, states that the available data do not allow a treatment recommendation to be defined in this indication and that further studies are needed. [Ref. 6] His 2020 intralipid series, of which he is 18th and last named author, covered 187 women treated between 2015 and 2018 in three university hospitals, of whom 26 received intralipid; among the ten who had recurrent miscarriage, with a median of five previous losses, seven pregnancies on intralipid resulted in a live birth, which was more frequent than in a comparison group of 20 untreated women, p = 0.02. [Ref. 7] On TNF alpha antagonists, he has published only two illustrative cases, in 2019, of which he is first author. [Ref. 8] His most recent work on the subject is a 2026 case report, of which he is 6th and last named author and the corresponding author: a 41-year-old woman with ten pregnancy losses and massive perivillous fibrinoid deposition carried a pregnancy to term on a combination of intravenous immunoglobulins, corticosteroids, hydroxychloroquine and antithrombotic treatment; the abstract of the article itself states that no agent can be assigned causality, because of the single case, the incomplete histological documentation of the previous losses and the simultaneous use of several treatments. [Ref. 11]

The FaCIL-2, NCT03970954 study explored the use of low-dose interleukin-2 as a strategy for modulating regulatory T cells in unexplained early miscarriage. It is a proof-of-concept study; the results available were released as a preprint and need confirmation in controlled studies.

Phase 1/2, sponsored by AP-HP, it has been completed since 20 March 2024, with 18 participants included. The registry declares Prof. David Klatzmann as principal investigator, and Arsène Mekinian as study director for the Saint-Antoine Hospital site, alongside Prof. Gilles Kayem for Trousseau Hospital. Arsène Mekinian is 1st of the eight named authors of the preprint reporting its results. Those results are a preprint, not peer-reviewed: they were posted on medRxiv on 16 June 2025 and otherwise exist only as a conference abstract published in June 2026 in a supplement of La Revue de médecine interne, volume 47, supplement 1, page A49, which is not an original peer-reviewed article. What the preprint reports: 15 women with at least five unexplained early pregnancy losses received 3 million international units of interleukin-2 per day for five days per cycle, from the tenth day after the onset of menstruation, with up to five cycles in the absence of pregnancy. The primary endpoint, an immunological one, was met, with a mean 2.0-fold increase in circulating regulatory T cells at day eight, p below 0.001, and an elevation still present at day 29. Of the eight pregnancies obtained, four were ongoing at twelve weeks, resulting in three live births and one late miscarriage at twenty weeks. Nine further patients of comparable profile, treated outside the trial on compassionate grounds and without immunological monitoring, had five pregnancies and two live births. The authors present this work as an open-label proof-of-concept study, without a control group, and conclude that a controlled trial with prolonged treatment in early pregnancy is needed. The declaration of interests states that two of the authors are inventors on patent applications covering the therapeutic use of low-dose interleukin-2. [Ref. 12]

Two recent reviews that he co-authored set out the state of knowledge as his team states it. The 2025 review devoted to hydroxychloroquine in recurrent immune-mediated obstetric disease, of which he is 5th and last named author, writes that the use of hydroxychloroquine has been empirical in unexplained recurrent pregnancy loss and that current data do not support its use in this indication; it also recalls that no randomised trial has assessed the addition of hydroxychloroquine to conventional treatment of refractory obstetric antiphospholipid syndrome. [Ref. 9] The 2026 review on the immunology of fertilisation, implantation and pregnancy, of which he is 4th of the five named authors, concludes that the efficacy of the immunomodulatory treatments explored, corticosteroids, intravenous immunoglobulins and intralipid, remains to be confirmed by large studies. [Ref. 10]

The aim is now to move from empirical immunomodulation to strategies assessed prospectively, based on a precise characterisation of patients and on clinically relevant obstetric endpoints.

Multidisciplinary care

Internal medicine and clinical immunology expertise is developed at the department of internal medicine and clinical immunology at Saint-Antoine Hospital, AP-HP – Sorbonne Université, in collaboration with the obstetrics and gynaecology and reproductive medicine teams.

Arsène Mekinian practises in the department of internal medicine at Saint-Antoine Hospital (AP-HP), attached to the medical and university department devoted to inflammation, immunopathology and biotherapies, designated DHU i2B until about 2019, then DMU 3iD and DMU i3 in the affiliations of his publications, at Sorbonne Université, working with the obstetrics and gynaecology and reproductive medicine teams. This is the affiliation that appears on all his publications on this subject, and it is the site declared to the registry for the FALCO cohort.

Women followed in this setting may be included in the FALCO national registry, whose public entry NCT05557201 declares Saint-Antoine Hospital as the recruitment site, whereas the 2024 publication names Saint-Antoine Hospital, Tenon Hospital and Trousseau Hospital. Repeated implantation failure falls, within the same registry, under the same inclusion criteria.

Complex cases may also be presented to the FAI²R network national Infertility multidisciplinary team meeting, which brings together obstetrician-gynaecologists, internists and pathologists.

Teaching in the field is organised at Sorbonne Université in the form of a university diploma in reproductive immunology, lasting seventy hours, entirely at a distance, of which the Sorbonne Université Continuing Education page, updated on 28 July 2026, states that Arsène Mekinian is the academic lead. The published calendar covers ten sessions, from October 2026 to June 2027.

What remains to be shown

That the immune system takes part in the physiology of implantation and pregnancy is established. Its causal role in unexplained miscarriage, on the other hand, and the possibility of identifying precisely the patients concerned, remain research questions.

To date, no single immunological biomarker allows the patients likely to benefit from immunomodulation to be selected in a sufficiently validated way. This is the result of his own work of 2019: the proportions of circulating NK cells and of T-cell large granular lymphocytes measured before conception distinguished neither the cases from the controls, nor the pregnancies that succeeded from those that ended, and the article concludes that these measurements should not serve as an indicator for management. [Ref. 3] The result of his 2021 cohort points the same way, none of the immune markers measured differing according to outcome. [Ref. 2] The work on the endometrial immune profile, which is more favourable, is retrospective, covers 104 patients, and the test it assesses is covered by a patent held by one of its authors, an employee of the company that markets it along with two other co-authors. [Ref. 4]

Hydroxychloroquine has not shown a benefit in this indication, and it is his own registry that establishes this. Of 100 exposed pregnancies, 62% ended before twelve weeks and 31.3% resulted in a live birth; the authors, of whom he is the last, write that this exposure does not appear to prevent further miscarriages. That cohort had no control group, a limitation they state first, and the only predictive factor found is not a treatment but the number of previous losses. [Ref. 1] The 2025 review he co-authored writes that current data do not support the use of hydroxychloroquine in unexplained recurrent pregnancy loss. [Ref. 9]

Results from observational cohorts, small series or uncontrolled studies must be distinguished from the evidence provided by randomised trials. The weightiest point concerns corticosteroids, because it sets an observation against a trial. His 2021 cohort reports that pregnancies treated with corticosteroids were twice as likely to result in a live birth as untreated pregnancies, odds ratio 2.0, interval 1.1 to 3.7. [Ref. 2] Yet a randomised double-blind placebo-controlled trial, conducted in 715 women in eight fertility centres in China and published in JAMA in 2023, showed no benefit of 10 mg of prednisone per day on live birth: 37.8%, that is 135 women of 357, in the prednisone group against 38.8%, that is 139 of 358, in the placebo group, absolute difference -1.0% with a 95% confidence interval of -8.1% to 6.1%, relative risk 0.97, interval 0.81 to 1.17, p = 0.78. That trial also observed more biochemical pregnancy losses on prednisone, 17.3% against 9.9%, relative risk 1.75, interval 1.03 to 2.99, p = 0.04, and more preterm deliveries, 11.8% against 5.5%, relative risk 2.14, interval 1.00 to 4.58, p = 0.04. That trial is not his: it is signed by a Chinese team, Sun Y et al., and Arsène Mekinian is not an author of it. Nor is its population exactly the same, since it concerns women with repeated implantation failure in in vitro fertilisation and not spontaneous recurrent early pregnancy loss. It therefore does not directly refute the result of the French cohort. But it is the only large randomised trial available on a corticosteroid in reproductive failure, its result is negative on live birth and it signals a risk, and it bears on the very principle of immunomodulatory treatment in these situations. An association observed in a non-randomised cohort does not carry the same weight as that trial, and this page does not claim otherwise.

The results of the FaCIL-2 trial are not peer-reviewed. They are available as a preprint on medRxiv since 16 June 2025 and as a conference abstract in a journal supplement in June 2026, which does not constitute peer review. The trial was open-label, without a control group, and covered 15 women treated, with three live births. Its primary endpoint was immunological and not obstetric. Its own authors conclude that a controlled trial is needed. [Ref. 12]

The remaining treatments cited in this field rest, within his body of work, on small open series, on non-randomised comparison groups of small size, such as the 20 untreated women of the intralipid series, or on isolated cases, such as the two TNF alpha antagonist cases of 2019 and the single case of 2026, whose abstract itself recalls that no causality can be assigned to any one agent. [Ref. 7, 8, 11] None of this work has been replicated in a controlled trial.

Two methodological difficulties run through the whole field and make these limitations structural. The first is the heterogeneity of definitions, two or three losses depending on the learned society, which prevents series from being compared with each other. The second is that a substantial proportion of women achieve an ongoing pregnancy at the next attempt with no particular treatment: a series without a control group, even with a high live birth rate, therefore does not allow the conclusion that a treatment works. This is the reason why the FALCO registry and the CERTIFY trial were built, and it is what the articles cited above say themselves.

Finally, the purpose of the FALCO registry is to describe and to compare, not to settle: to date it has produced only one publication, that of 2024, and its primary data collection is not complete. The priorities are therefore to develop reproducible phenotypes, validated biomarkers and controlled therapeutic trials using in particular live birth as the clinically relevant endpoint.

References

Published work he has taken part in, on this subject.

General information only. This page does not replace individual medical advice. Updated: 19 September 2026.