Azoospermia is a highly adverse finding in the male partners of infertile couples. Although a minority of azoospermic men may benefit from surgical interventions and medical therapy, eventually leading to an emergence of sperm in the seminal fluid, most of them are naturally considered sterile and depend on sperm retrieval techniques in order to achieve biological fatherhood. However, successful surgical sperm retrieval may be guaranteed only in cases of obstructive azoospermia. In contrast, in non-obstructive azoospermia (NOA), which is characterized by spermatogenic failure, the sperm retrieval rate (SRR) tends to be modest at best. The SRR in NOA essentially depends on the surgical technique and testicular phenotype. Microdissection testicular sperm extraction (microTESE), originally proposed by Schlegel and Li [1], is the most appropriate technique for NOA. As for the testicular phenotype, there is no way of properly determining it before testicular biopsy per se, which substantially limits our understanding of the likelihood of sperm retrieval before attempting microTESE and makes patient counseling suboptimal. Even diagnostic testicular biopsy, as demonstrated by Berookhim et al. [2], has little predictive value due to the potential heterogeneity of seminiferous tubules and the random nature of this procedure.

Endocrine parameters and testicular volume are currently viewed as possible markers of microTESE success. Follicle-stimulating hormone (FSH), despite being routinely used by clinicians in decision-making along with inhibin B, seems to be quite unreliable, with a poor area under the curve (AUC) value demonstrated in a meta-analysis based on five studies [3]. Even if we consider it to be a predictor for successful sperm retrieval, it is definitely not a linear one. Indeed, a weak U-shaped relationship between FSH level and the SRR was noted in at least 1 study [4].

It would be more logical to try to integrate these clinical and biochemical parameters together in order to identify basic patterns that correspond to distinct testicular phenotypes in NOA. In other words, the likelihood for successful sperm retrieval does not and should not depend on individual endocrine parameters. Rather, using endocrine parameters as a whole, we may deduce the current intrinsic state of testicular parenchyma, and the latter effectively defines the SRR. The evaluation of clinical data, hormone levels, and testicular pathology is currently disjointed. The widely used classification, which includes hypospermatogenesis, maturation arrest, and Sertoli cell-only (SCO) syndrome, does not reflect all the complexities that underlie these phenomena.

We aimed to perform a heuristic identification of basic testicular phenotypes by comparing observed clinical patterns with pathology results and reproductive outcomes. An often overlooked clinical pattern is isolated FSH elevation (iFSH) in the presence of normal or borderline luteinizing hormone (LH) and testosterone levels. This finding is quite frequent in NOA, but it is not as widely discussed as more obvious and well-known patterns, such as hypergonadotropic hypogonadism. Therefore, we decided to begin with an attempt to describe iFSH, which seems to be the most unfavorable endocrine pattern, as pointed out by Esteves et al. [5]. Our hypothesis was that iFSH may reflect a specific testicular phenotype characterized by a negligible SRR and adverse pathological findings.

FSH is one of the main hormonal factors that regulate spermatogenesis. Its biological actions are mediated through receptors located on the surface of Sertoli cells [9]. It is well known that mutations affecting FSH or FSH receptor structure may have deleterious impacts on male fertility, although such cases are rarely observed in clinical practice [10,11]. Although the idea that iFSH-associated azoospermia has very poor SRR outcomes is not essentially new, we hypothesize that it represents the most adverse phenotype in the azoospermia landscape [5].

In research and clinical practice, the FSH level is treated as a serum marker for sperm retrieval failure, but this premise is biased. There is a weak, if any, linear association between the numerical value of serum FSH concentration and the likelihood of preserved spermatogenesis. The pooled analysis by Yang et al. [12] revealed an AUC of 0.72±0.04 in a ROC curve analysis of FSH as a sperm retrieval predictor; this AUC is moderate at best and reflects the inadequacy of FSH as a standalone marker. In a recent study by Zeadna et al. [13], FSH was likewise not found to be a valuable predictor of the SSR. In a study by Kelsey et al. [14], the FSH level had a fair AUC of 0.89 in simply distinguishing azoospermic and non-azoospermic childhood cancer survivors; however, with specificity and sensitivity of 81% and 83%, respectively, in this setting it is doubtful that the FSH level would be able to solve the significantly more delicate task of distinguishing azoospermic patients with focal spermatogenesis. Controversially, some authors have even described a positive association between FSH and the SRR [15]. We assume that this discrepancy can be resolved if we do not interpret the FSH level as a continuous variable, but rather as a flag of certain testicular phenotypes, such as the one we attempted to describe in this paper.

Interestingly, FSH levels in patients with iFSH-associated azoospermia tend to be only mildly or moderately elevated. This pattern can be explained by the involvement of a large number of factors in hypothalamic-pituitary-gonadal (HPG) regulation. Severe disruption of spermatogenesis is characterized by low levels of inhibin B, which means that this negative feedback loop is inactive, leading to uninhibited LH and FSH production. However, when Leydig cells are functioning normally, testosterone and estradiol exert proper negative feedback, which limits gonadotropin release so that LH remains borderline normal and FSH remains mildly elevated. This difference is due to the different effects of gonadotropin-releasing hormone (GnRH) on their production; in short, FSH seems to be less dependent on GnRH pulse frequency and seems to have a basal constitutive secretion [16]. Furthermore, inhibin B directly blocks activin-induced FSH secretion by pituitary gonadotropes, while its influence on LH secretion is limited [17]. This scenario is possible in cases of germ cell aplasia or, possibly, selective death of germ cells due to viral infection or another testicular insult that leaves interstitial testosterone-producing cells intact. In contrast with iFSH-associated azoospermia, in complete testicular failure with impaired spermatogenesis and sex steroid synthesis, both negative feedback loops are broken, which results in full-blown hypergonadotropic hypogonadism.

As for the difference in the SRR, germ cell aplasia or precise depopulation due to currently unrecognized factors would lead to a virtually zero chance of sperm retrieval (iFSH scenario). However, gross damage to the testicular parenchyma may actually leave some tubules relatively intact, with evidence of their continuing function being masked by the overwhelming HPG response when two major loops of negative feedback are severed (the hypergonadotropic hypogonadism scenario). In fact, it is the latter scenario in which microTESE technique works best. A study by Yu et al. [18] paradoxically demonstrated that patients with narrower seminiferous tubules and higher FSH levels had better chances of successful microTESE. Their findings may actually reflect the same phenomenon and even provide some indirect morphological proof. The relatively good predictive value of FSH for sperm retrieval in conventional TESE (cTESE), as observed by Gnessi et al. [19], could be a result of bias. Patients with hypergonadotropic hypogonadism and focal spermatogenesis benefit less from cTESE than they would benefit from microTESE, which negates the principal difference between them and patients with iFSH. This results in a more uniform SRR and similarly poor outcomes in all FSH groups. Some successful cTESE attempts in mildly hypergonadotropic patients with hypospermatogenesis would be possible, which would give the appearance of a “classic” inverse relationship between FSH and the SRR. Indeed, the aforementioned meta-analysis by Yang et al. [12] showed a reasonable AUC of 0.72 for FSH, but it included studies on both cTESE and microTESE. The same logic would apply to testicular sperm aspiration (TESA), which is essentially random, like cTESE, while being less invasive. A recent study by Liu et al. [20] revealed a negative correlation between the FSH level and the SRR for TESA, as expected; however, there was a positive correlation between the FSH level and microTESE success. In a study by Zhu et al. [21], FSH had an AUC of 0.87 for successful surgical sperm retrieval, but the majority of patients underwent simple cTESE. Nonetheless, the meta-analysis by Li et al. [3], based strictly on microTESE studies, demonstrated an abysmal AUC of 0.612. The AUC in our series (0.704) is well within the range of values observed by other researchers.

In microTESE, a U-shaped association between FSH and the SRR was described. The concept of iFSH-associated azoospermia fits this concept perfectly. In these adverse cases, FSH tends to be moderately elevated, which forms a “low SRR valley” in the U-shaped FSH-SRR curve. Ramasamy et al. [4] observed this “valley” with an SRR of 51% at FSH levels less than 15 IU/mL, while the SRR in other groups was over 60%. In another study, the worst prognosis for sperm retrieval was observed for FSH levels within the 10–15 IU/mL range, though it described a specific subpopulation of patients with SCO histology on diagnostic testicular biopsy [2]. Zhang et al. [22] also demonstrated poor microTESE outcomes with a “low SRR valley” in patients with moderate elevation of FSH, although only when the testicular volume was low. The authors [22] analyzed FSH and LH levels separately, so it is unknown whether their results could be influenced by the iFSH pattern. If the “low SRR valley” hypothesis is true, it invalidates the concept of an “FSH cut-off point” to select patients in whom microTESE is allegedly pointless. However, Chen et al. [23] reportedly found that an FSH cut-off point of 19.4 IU/L predicted the absence of testicular sperm, although their study was based only on a direct comparison of average FSH levels in patients with positive and negative testicular biopsies. The mean reported FSH level in patients with successful sperm retrieval was only 7.94±4.95 IU/L, which hints at the possible contamination of this group by obstructive and transitory azoospermia cases.

Most iFSH patients had idiopathic azoospermia, which is likely due to unrecognized mutations that are deleterious for germ cells. A recent study by Das et al. [24] is of interest in this context, as idiopathic NOA patients had significantly lower (though still out of reference range) levels of FSH than patients with a known etiology of NOA. The authors [24] did not report LH levels in their cohort, so it is not possible to speculate whether their patients had iFSH. However, considering that testosterone levels were higher in idiopathic NOA patients, this might be the case. We also had seven patients who previously underwent cytotoxic cancer treatment, which, as a side effect, possibly eradicated the germ cell epithelium. Norwegian researchers specifically described this as a possible adverse event in patients treated for malignant lymphomas, naming it “exocrine hypogonadism” [25]. The FSH elevation observed in human males with age may also be related to a gradual depletion of germ cells [26]. In theory, any man who would live indefinitely long would eventually develop iFSH-elevated azoospermia.

Our study is limited by its retrospective nature, lack of an adequate control group, and the fact that iFSH definition was based on local laboratory reference values. FSH production depends on many factors beyond feedback, including the general responsiveness of the HPG axis, gonadotrope population, GnRH pulse frequency, and basal FSH production. Laboratory reference values are not universal in a biological sense, and even while cases with a moderately elevated FSH level and a borderline elevated LH level could potentially bear the same iFSH-associated phenotype, they were not considered to have iFSH due to the reliance on reference values. Therefore, results obtained in this study should not be generalized, but could be used to inform further research.

In this paper, we made an attempt to describe isolated FSH elevation as a distinct clinical pattern in patients with azoospermia, and also frequently mentioned hypergonadotropic hypogonadism as another typical pattern in azoospermic patients, mainly as a reference point for comparisons. Other specific phenotypes of NOA may exist, and they need to be properly described. In our opinion, numerical variables that reflect endocrine parameters are easy to analyze and serve as convenient candidate markers of reproductive outcomes and nomogram material, but we need to move on to recognize non-linear associations and patterns in NOA.