Across 2,054 men with unexplained infertility, a genetic cause was found in 11 per cent — and ten genes were associated with very low pregnancy rates after assisted reproduction.

A man with no sperm or very few, a normal karyotype and no Y chromosome deletion is told his infertility is idiopathic, which means the available tests found nothing. That category is large, and it is where genetics has the most room to work.

A study in Science Bulletin, led by researchers at Central South University and the Chinese University of Hong Kong, exome-sequenced 2,054 such men and then did something most genetic studies of infertility do not: followed what happened when they and their partners went through assisted reproduction.

A genetic diagnosis was found in 217 men, or 10.6 per cent. Counting cases supported by natural knockouts in other species raised that to 235, or 11.4 per cent.

Where the diagnoses came from

The genetic architecture turned out to be concentrated at one end and very diffuse at the other. Fifteen genes accounted for nearly half of all diagnoses, with DNAH1, MEI1 and GCNA the most common — genes involved respectively in building the sperm flagellum, in meiosis, and in protecting the integrity of germ cell genomes.

The rest is a long tail, which is the characteristic shape of male infertility genetics. Spermatogenesis requires an unusually large number of genes, many of them used nowhere else in the body, and a loss-of-function variant in any one of them can stop the process — which means thousands of rare causes rather than a handful of common ones.

Thirteen genes with prior evidence only from model organisms added 0.9 per cent to the diagnostic yield, and the study reports them as never previously documented in human infertility. That is a small number carrying a disproportionate amount of method: a mouse knockout that cannot make sperm is a hypothesis, and finding the human equivalent in a patient converts it into a diagnosis.

The team also identified 41 further candidate genes that may cause infertility.

Diagnostic yield was higher in men with oligozoospermia — few sperm — than in men with non-obstructive azoospermia, who have none. That is worth noting because it runs against the intuition that the more severe phenotype should have the more findable cause.

The part that is new

The clinical contribution is the outcome data, and the headline result is a reassuring null.

Among 905 men with non-obstructive azoospermia, sperm retrieval rates did not differ significantly between those with a genetic diagnosis and those without. Overall success rates in assisted reproduction were similar across groups.

That matters because the alternative result would have been worse. A genetic diagnosis that systematically predicted failure would turn a test into a reason to deny treatment, and this finding says that on average it does not.

The exceptions are specific. Men with defects in ten particular genes had very low pregnancy rates, and in five of those genes no live births occurred at all. DNAH1 was linked to lower live-birth rates, and loss of function in C12orf40 demonstrated something clinicians see occasionally and cannot usually explain: that ejaculated sperm are no guarantee that intracytoplasmic sperm injection will work.

“By pairing large-scale genomic profiling with ART outcomes, we provide actionable insights that can inform whether to pursue testicular sperm retrieval, use donor sperm, or anticipate ICSI prognosis,” said Yue-Qiu Tan of Central South University, the corresponding author. “This work supports integrating genetic testing into routine evaluation to personalize care for infertile men.”

“Patients and clinicians often face uncertain decisions about biopsy, ICSI, or using donor sperm,” said Zirui Dong of CUHK, a co-corresponding author. “Our gene-level prognosis data begin to clarify where success is more or less likely, helping set realistic expectations.”

How much weight the gene-level numbers can carry

The phrase to hold onto is Dong’s “begin to clarify”, because the authors’ own stated limitations bear directly on the most quotable finding.

They note that the analysis focused on the first ART cycle with few cases per gene. In a cohort where 235 men received a diagnosis spread across dozens of genes, the number of patients behind any individual gene is small — and “no live births” from a handful of cycles is a different statement from a zero success rate. Five genes with no live births could be five genes that genuinely preclude success, or genes with three or four attempts each and ordinary bad luck.

This is the standard difficulty with gene-level prognosis, and it is why these findings function as hypotheses for counselling rather than as probabilities to quote to a patient. A couple told that a variant means no chance is being told something the data cannot yet support.

The authors also flag that outcomes were not adjusted for female factors, which is the largest confounder available. Female age dominates ART success more than almost any other variable, and comparing live-birth rates between groups of men without accounting for the ages and fertility of their partners leaves open how much of any difference belongs to the man.

Cohort composition bias is their third caveat: a tertiary fertility centre in China sees a particular population, and diagnostic yields vary with ancestry, referral patterns and how strictly idiopathic was defined.

And exome sequencing reads the protein-coding fraction of the genome. It cannot see variants in non-coding regulatory regions or epigenetic changes, both of which are plausible contributors to spermatogenic failure — so the 88 per cent who remained undiagnosed are not necessarily men without a genetic cause.

What a diagnosis is for

The practical value of a result like this is not usually treatment, since no genetic cause of spermatogenic failure is currently correctable. It is information.

A man who learns the cause of his infertility gains an explanation where there was none, which has its own worth. More concretely, a diagnosis can indicate whether testicular sperm retrieval is likely to find anything before an invasive procedure is attempted, can identify variants that may be passed to a son conceived by ICSI, and can occasionally flag a gene with consequences beyond fertility.

Set against that is the cost of a wrong inference. Prognostic genetics in reproduction operates where people make irreversible decisions under time pressure, and a figure derived from four patients does not belong in that conversation as a number. Treated as what it is — a signal that some genes warrant closer attention and more data — the study moves the field forward without that risk.

The paper was published on 4 August 2026. The release does not list the paper’s authors beyond the two corresponding authors, does not state funding sources, and states no conflicts of interest.

---

Notes

1. Science Bulletin, 2026. DOI: 10.1016/j.scib.2026.08.003. The release does not give the paper’s title or author list.
2. Science China Press. “Does genetics in infertile men impact on the pregnancy outcome through assisted reproductive technology？.” EurekAlert!, 4 August 2026. https://www.eurekalert.org/news-releases/1138746.