The male sex chromosome is particular for its size, genomic structure, content, and evolutionary trajectory [44,45]. It is haploid, and it precludes recombination with the X chromosome for most of its length (except the two pseudoautosomal regions, PAR1 and PAR2). Moreover, it contains segmental duplications arranged in direct or inverted repeats, and palindromes [44]. The existence of these duplicated sequences allows a mechanism called Non-Allelic Homologous Recombination (NAHR), which might lead to recurrent deletions/duplications affecting the gene dosage on the Y chromosome. The first de novo deletions on the long arm of the Y chromosome (Yq) were described by Tiepolo and Zuffardi in 1976 [46], predicting the presence of Azoospermia Factor(s) (AZF). Further research defined three deletion patterns in proximal, middle and distal Yq11, designated as AZFa, AZFb and AZFc [47]. AZF microdeletions are generally de novo and their origin most likely derives in the testis of the patient’s father. In fact, during meiosis NAHR between sister chromatids may take place leading to AZF deletions [48]. Five Yq fragile sites exist, resulting in the recurrent removal of DNA segments ranging from 0.8 to 7.7 Mb. The most frequently deleted subregion is the AZFc (around 80%), followed by AZFa (around 0.5–4%), AZFb (around 1–5%), and AZFbc, with two different breakpoints (around 1–3%) [49]. The AZF microdeletions are well-known causes of male infertility since AZF regions comprise important spermatogenesis associated genes and gene families [49]. The frequency of complete AZF deletions is 1:4000 in the general population, but it rises up to 5–10% of patients affected by idiopathic NOA [49,50]. The deletion phenotypes for each region are reported in Figure 2. Due to the presence of several genes and their multicopy nature in these regions, it is difficult to understand which genes cause the associated phenotype.

The AZFa region is 792 kb long and contains two ubiquitously expressed, single-copy genes, USP9Y and DDX3Y (former DBY). USP9Y encodes a protein with ubiquitin C-terminal hydrolase activity, which may play an important regulatory role at the level of protein turnover [51]. The isolated absence of this gene is associated with a large spectrum of testis phenotypes, ranging from azoospermia with hypospermatogenesis to normozoospermia [52]. Based on this, USP9Y is more likely to be a fine tuner that improves efficiency, than a gene with an essential function. On the other hand, DDX3Y encodes an ATP-dependent RNA helicase that is a member of the well-conserved DDX3 DEAD Box Helicase family [53]. DDX3Y protein was found predominantly in spermatogonia, whereas its X chromosome homologue, DDX3X, was found to be expressed after meiosis in spermatids. Although isolated mutation or removal of the DDX3Y gene has not been reported, it is likely that the removal of DDX3Y gene is responsible for the AZFa deletion phenotype, which is SCOS. SCOS is characterized by the total absence of germ cells in the testis, low testis volume and high FSH.

Complete AZFb deletion removes a 6.2 Mb DNA segment, including 32 copies of genes and transcription units. These genes are likely to be involved in germ cell maturation since their removal causes spermatogenic arrest. MA is a cessation of germ cell formation mainly at the spermatocyte stage, resulting in azoospermia. MA is typically associated with normal testicular volume and normal gonadotropin levels (LH, FSH), which might mimic obstructive-azoospermia.

The AZFc region involves 12 genes and transcription units, each present in a variable number of copies resulting a total of 32 copies. The clinical manifestation in complete AZFc deletion carriers is largely variable. Spermatozoa may be detected in the ejaculate but typically less than 2 million/mL [50]. Since a progressive decline in sperm count has been observed among these subjects, sperm cryopreservation should be offered to prevent future testis surgery. In those patients who present azoospermia the testis phenotype ranges from SCOS to hypospermatogenesis.

Y chromosome deletion screening is performed in a standardized way, according to the EAA/EMQN guidelines [6]. Two markers are used for each AZF regions besides the control markers of SRY and ZFX/ZFY. At this regard, it is worth to note that one of the two AZFa markers proposed in the guidelines, contains a polymorphic site in sY84 primer sequence [54]. This SNP (rs72609647) is relatively frequent in the Chinese population while it is yet undetected in the Caucasian population. It is therefore mandatory to use alternative STSs in case of failed amplification of the sY84 marker. For neighboring markers, the “MSY breakpoint mapper” can be consulted (http://breakpointmapper.wi.mit.edu/) [6].

The AZF deletion screening does not only useful for diagnostic purposes but it is also important for TESE prognosis (Figure 2). Carriers of complete AZFa and AZFb deletions have virtually zero chance to find spermatozoa in their testes. In fact, the correct distinction between complete and partial AZFa or AZFb deletions is mandatory through the second step, extension analysis of the AZF deletions. Concerning the AZFb region, Stouffs and colleagues [55] reported two patients with suspected complete AZFb deletion based on the routine first step analysis but with residual sperm production. By performing the extension analysis of the deletion in these patients, the authors were able to amplify a specific distal STS (sY1192) of the AZFb region. This finding indicates that the retention of some copies of multicopy genes such as PRY, RBBMY, BPY2, DAZ and some transcription unites may allow residual spermatogenesis and complete maturation. From a clinical point of view, this report called the attention to the importance of testing for sY1192, which is now considered as a decision-making marker: if it is present, TESE might be attempted (Figure 2).

Concerning AZFc microdeletion carriers, spermatozoa can be retrieved through m-TESE from the testicles with a success rate of 50–60% [56]. Data in the literature indicate that a significant proportion of spermatozoa from AZFc microdeletion carriers are nullisomic for sex chromosomes [57,58]. Yq microdeletions could be associated with an overall Y chromosomal instability, which can lead to the formation of 45,X0 cell lines. Karyotype analysis with the search of 45,X0/46,XY mosaicism should be performed on peripheral blood lymphocytes, since mosaicism is considered as a negative predictor of sperm retrieval success [57,59,60,61].

CHH is a rare endocrine disease (1:8000 males), caused by the deficient production, secretion or action of the Gonadotropin-Releasing Hormone (GnRH), in the absence of anatomical or functional abnormalities of the hypothalamic-pituitary axis. CHH is a clinically heterogeneous condition covering a wide spectrum of symptoms, where the typical clinical features are delayed puberty and azoospermia. CHH can manifest itself with anosmia/hyposmia (Kallmann Syndrome) or as a normosmic form (nCHH). In addition, non-reproductive features can be recognized in CHH patients, i.e., midline facial defects (cleft lip or palate), unilateral renal agenesis (URA), hearing loss, synkinesia, dental agenesis and short metacarpals [62]. Furthermore, CHH can occur as a part of complex genetic syndromes, such as CHARGE syndrome, Gordon Holmes syndrome and Waardenburg syndrome. CHH is heterogeneous not only clinically but also genetically. To date, more than 40 genes with variable expressivity, penetrance and inheritance have been identified as the genetic cause of the disease and reviewed in two recent articles [7,8]. Currently, through the sequencing of a targeted-gene panel using NGS, a genetic diagnosis is possible in about 50% of cases, and it is expected that novel CHH-associated genes will be discovered by WES analysis in the near future. To note, the main challenge in NGS –based methods is the interpretation of variants of unknown significance (VUSs), hence appropriate tools and expertise are needed for correct interpretation of these findings in the clinical practice.

CHH presents three peculiar features: (1) distinguishing between KS and nCHH is often difficult from a genetic point of view, because mutations in genes involved in GnRH-mediated neuronal migration might result in both forms of CHH [63]. With the exception of certain genes purely associated with Kallmann syndrome or with nCHH, some others (for instance FGFR1 and PROKR2) can be involved in both clinical manifestations, even within the same kindred [7]. (2) CHH is no longer viewed as a Mendelian monogenic disease, since rare variants in two (digenicity) or more (oligogenicity) candidate genes have been found in the same patient, supporting a digenic/oligogenic inheritance in about 20% of cases [63]. Although the oligogenic basis of CHH makes the genotype-phenotype correlation even more complex, it may explain the variable penetrance of the same pathogenic variant within the same family members [7,8]. (3) The traditional view of CHH as a lifelong disease has been changed following the observation of spontaneous remission in patients affected by Kallmann syndrome or nCHH, regardless of type of identified genetic defect [64]. Thus, a periodic suspension of the substitutive testosterone therapy is advised in order to verify the “reversibility”.

In about 80% of CHH patients, spermatogenesis is induced after 9–18 months of gonadotropin treatment [64], and mutations can be transmitted either through spontaneous pregnancy or through Assisted Reproductive Techniques (ART). Overall, the complexity of this disease makes predicting the exact health consequences for the offspring difficult; however, PGD or prenatal diagnosis should be offered to couples, only for syndromic cases and after taking into consideration national legislations.

CBAVD is a congenital developmental disease (1:1000 males) characterized by the lack of both vas deferens. It may manifest with various clinical features depending on the association or not with other abnormalities of the male urogenital tract, mainly of the seminal vesicles (50% of cases) and the kidneys (renal agenesis occurring in 5–10% of CBAVD patients) [65]. The prevalence of CBAVD in azoospermic men is estimated to be 4–17% and raises up to 25% in case of OA [3,66]. CBAVD associated with agenesis of the seminal vesicles is characterized by a typical semen phenotype, consisting of azoospermia, low semen volume (<1 mL), acid pH (<7). In contrast to CHH, which is a genetically heterogeneous condition (see above), the genetic contribution of CBAVD without kidney anomalies is limited to two genes: (1) Cystic Fibrosis Transmembrane Conductance Regulator (CFTR) gene for the 75–80% of cases and (2) Adhesion G Protein-Coupled Receptor G2 (ADGRG2) for the 5–10% of cases.

CFTR gene spans 250 kb on the long arm of chromosome 7 (7q31.2), including 27 exons, and encodes for a functional protein of 1480 amino acids. The CFTR protein is involved in the regulation of several ions transporters, including sodium channel, chloride/bicarbonate exchangers, proton exchangers and water channels, and its biallelic dysfunction is responsible for the Cystic Fibrosis (CF) phenotype. Up to now, more than 2000 mutations have been reported in this gene (http://www.genet.sickkids.on.ca/). Disease causing mutations in CFTR may affect not only coding regions but also the promoter and deep intronic regions [67,68]. CFTR disease-causing alleles can be classified into two main types: (i) “severe” or CF-causing mutations, which are always associated with CF in a homozygous state; (ii) “mild” or non-CF-causing mutations, which have never been observed in CF patients. The presence of two “mild” mutations or one “severe” plus one “mild” allele is causative for CBAVD. Up to 33% of CBAVD subjects of European descendent are compound heterozygous carrying the CF-causing mutation F508del (c.1521_1523delCTT; p.Phe508del) and the non-CF-causing intronic 5T allele (IVS8-5T allele), whose frequency is 4-to-5 times higher in CBAVD patients [69]. However, it is worth noting that some CFTR variants are associated with variable expressivity, also called as “variants of varying clinical consequence”, for instance R117H (c.350G > A; p.Arg117His) and IVS8-5T allele. It remains difficult to predict the associated phenotypes in those cases in which one of these variants is in trans with another one of the same type or when it is in trans with a known pathogenic variant [70]. Thus, a genotype involving these variants should be interpreted carefully and reported thoughtfully. For instance, the penetrance of 5T allele is dependent on the status of the adjacent poly TG tract, which usually contains 11, 12, or 13 repeats (c.1210–34TG [11], c.1210–34TG [12], c.1210–34TG [13]). When paired with a known CF-causing variant, 5T and 11TG variants in cis rarely predispose to CBAVD, whereas 5T in cis with 12TG or 13TG confers a risk for CBAVD. Given that 1:10 individuals carry the 5T variant, interpretation of its clinical value should always be performed in the context of the number of associated TG repeats [70].

ADGRG2 is an X-linked gene encoding an adhesion-class G protein-coupled receptor and is highly expressed in the efferent ducts [71]. In 2016, Patat and colleagues identified 3 hemizygous truncating mutations of the ADGRG2 gene in 4 out of 26 selected CBAVD subjects with normal kidneys and wild-type CFTR gene [72]. Interestingly, Adgrg2-mutant mice develop fluid accumulation in the deferent ducts, leading to an obstructive infertility phenotype, which resembles that observed in men with ADGRG2 mutation [73]. Three other studies have subsequently reported the identification of five new rare variants of ADGRG2 in 6 CBAVD patients of Asian origin with no pathogenic CFTR mutation [74,75,76]. Recently, six novel truncating mutations of ADGRG2 have been reported in a cohort of 53 French CAVD men carrying none or only one CFTR mutation and not presenting with URA [77]. The authors suggested that hemizygous ADGRG2 mutations are responsible for CBAVD phenotype with normal kidneys, accounting for approximately 20% of unexplained CBAVD cases after a comprehensive analysis of CFTR gene [77].

Despite this later discovery, 10–15% of CBAVDs remains without a genetic diagnosis. A portion of these unexplained CBAVD cases coexists with URA, suggesting an early organogenesis disorder. On the contrary, CBAVD related to CFTR or ADGRG2 mutations might be the result of a progressive degeneration that begins later in fetal life and probably continues after birth [65]. The small percentage of CBAVD cases with normal kidneys lacking of genetic diagnosis could be explained by defects in additional genes. For instance, the SLC9A3 gene has been proposed to be involved in some idiopathic CBAVD cases of Taiwanese origin [78,79]. In addition, epigenetic or environmental factors could contribute to the development of this disease [65].

The clinical management of men affected by CBAVD without renal agenesis must include CFTR analysis, which should be based on two steps: (1) a targeted panel of the most commonly known CF-causing mutations in the Caucasian ethnic group, including R117H and 5T allele; (2) a comprehensive screening of CFTR for rare point mutations and large rearrangements, when none or only one single mutation has been identified by the first step. If the CBAVD man has no biallelic mutations after comprehensive CFTR testing, the analysis of ADGRG2 is suggested, especially if there is a family history of male infertility. With the high accessibility to NGS-based testing, it could be more cost-effective to perform directly a comprehensive scanning of both genes [65]. Again, the main issue concerns the assignment of a clinical value to VUSs identified by NGS-based methods.

Since the testicular function of CBAVD patients is usually normal, conception of a biological child is possible through TESE combined with ICSI. Given that the carrier frequency of CFTR mutations in the Caucasian ethnicity is high (1:25), the CFTR panel screening in the female partners is mandatory. If mutations are detected in both partners, the risk of an offspring with CF (classical CF or non-classical CF) is very high and PGD should be advised to the couple.

NOA (1:100 males) represent >70% of total azoospermia cases and the majority of them shows primary testicular failure due to an intrinsic defect in the onset or progression of spermatogenesis. As reported in the introduction, NOA can manifest with various testicular histologies, including SCOS, MA at different stages and hypospermatogenesis. SCOS and complete MA are characterized by the absence of haploid cells in the testes, hence the TESE procedure to recover spermatozoa for subsequent ICSI is unsuccessful. In case of incomplete MA, some tubules containing round or later stage spermatids may be found. For this purpose, the development of a pre-TESE diagnostic gene panel would be of high clinical benefit to prevent unnecessary surgery in patients with pure SCOS/MA.

To date, the knowledge of monogenic causes of NOA due to primary testicular failure is limited and none of the current clinical guidelines includes mutational analysis of any NOA genes [9,11]. Given that the spermatogenic process is inherently complex and >2000 genes participate in it [80,81], a high genetic heterogeneity seems to be plausible. So far, 38 candidate genes for iNOA were reported based on the resembling reproductive phenotype in the mouse or on the family segregation (Table 1). Currently, only 17 of them have reached “moderate” or higher grade of clinical evidence, according to the classification criteria proposed by Oud and colleagues [9].

These genes are responsible either for isolated form of NOA or for complex phenotypes where azoospermia may represent one of the clinical symptoms underlying a certain syndrome. In this latter case, we can refer to them as “phenocopy” genes. It is worth noting that, for certain congenital diseases, syndromic features may be non-penetrant in the patient, as well as in the family, or may become evident only with age. For instance, in FANCA-mutated patients affected by iNOA, the complex Fanconi Anemia (FA) phenotype may be subtle and need to be specifically assessed to reach a diagnosis of “occult” FA in adulthood (see below; [82]). Thus, in case of pathogenic variants in “phenocopy” genes, a clinical re-evaluation of the patients and relatives should always be part of the diagnostic workup of NOA men.

Very recently, by using STRING analysis of physical and functional protein-protein interactions, Kasak and Laan demonstrated that candidate NOA genes belong to a dense network of “predicted functional partners” [11]. In particular, two distinct clusters of proteic interactions were observed, one for isolated NOA and the other one for syndromic conditions, emphasizing the different etiology underlying the two forms of NOA.

With the exception of TEX11 and WNK3 which are X-linked, the remaining candidate genes are mapping to autosomes. Apart from DMRT1, NR5A1, PLK4 and WT1 which follow the autosomal dominant inheritance pattern, only recessive mutations lead to quantitative alterations of spermatogenesis (Table 1).

In this paragraph, we will discuss the candidate genes for iNOA (Table 1) according to the different associated-testicular histologies such as SCOS, MA, different types of testis phenotype and undefined testis phenotype.