Genetic Prole of Inborn Errors of Immunity Using Whole Exome Sequencing in Individuals With BCG Localized Adverse Events

Purpose: In Mycobacterium tuberculosis endemic regions, BCG vaccine is administered early after birth to confer protection against severe form of tuberculosis disease. Previous reports suggest that BCG adverse events, even localized ones (BCGitis), can be the rst manifestation of immunodeciency. We investigated children with a history of BCGitis who needed drug treatment looking for possibly pathogenic variants in inborn errors of immunity genes (IEI-genes). Methods: Forty-four probands were evaluated. The exome sequences obtained by Next-Generation Sequencing were ltered for variants in the 344 IEI-genes described by the International Union of Immunological Societies (IUIS) and classied according to the recommendations of the American College of Medical Genetics. The identied candidate variants were validated by Sanger sequencing. Results: Out of the 44 probands, 36 were sporadic cases and 8 were familial cases. Thirty-one in 44 (70.5%) presented immunoallergic or other infectious clinical conditions besides BCGitis; 19 in 44 (43.2%) presented variants classied as pathogenic or likely pathogenic in 17 different IEI-genes, of which 35.3% were genes related to defects in intrinsic and innate immunity, including Mendelian Susceptibility to Mycobacterial Disease (MSMD) genes (IRF8, IFNGR1, JAK1, STAT1, TLR3 and TBK1). Remaining genes were distributed in another ve IUIS classications groups (CARD14, CFH, CHD7, FOXN1, NFAT5, NLRP3, NOD2, PMS2, STAT3, TNFRSF13B and TNFSF12). Conclusion: The high prevalence of pathogenic or likely pathogenic variants found in IEI-genes may be associated with BCGitis, which should be considered a sign of an inborn error of immunity.

relationship between related genes to Mendelian Susceptibility to Mycobacterial Disease genes (MSMD-Genes) to BCGitis, three newly described genes in the IUIS update [20] were added to our analysis (Table S1). Further ltering steps were applied as speci ed on supplemental methods. Candidate variants were validated by Sanger sequencing and then submitted to familial segregation.
To assess the importance of IEI-related genes on patients presenting with BCG adverse event, an enrichment analysis was performed using as control population WES data from 1,562 unrelated samples from the Baylor Hopkins Center for Mendelian Genomics, searching for rare single nucleotide variants (SNV) on IEI-related genes. A contingency table containing the number of individuals presenting these quali ed SNVs was accomplished for each gene found in the BCG cohort samples and in the control samples. The p value was determined by Fisher's exact test and it was subsequentially was corrected by Benjamini and Hochberg method.
For variant interpretation, the American College of Medical Genetics and the Association for Molecular Pathology (ACMG/AMP) guidelines [21] were used.
Probands who presented variants in IEI-Genes classi ed as Pathogenic or Likely Pathogenic were categorized according to clinical outcome in respect to the development of other symptoms besides BCGitis: Only BGCitis group, with probands who did not present other symptoms besides BCGitis; Mild or Transient Infectious Conditions group, with probands who presented mild or transient infectious conditions besides BCGitis; and Immunoallergic Conditions group, with probands with immunoallergic conditions besides BCGitis.

Ethics
This study was approved by the Ethics Committee of Universidade Federal de São Paulo (protocol number 842,006 and 1,641,734). All participants or legal guardians signed the consent forms.

Results
The 44 children (probands) with BCGitis from non-consanguineous Brazilian families included 36 probands without family history of other BCGitis (sporadic cases) and 8 probands with familial BCGitis (Figure 1, Table S2).
In four familial cases it was possible to analyze additional affected individuals (multiplex). In another four familial cases, only proband was analyzed (simplex). Among the probands simplex families, variants in IEI-Genes classi ed as Pathogenic or Likely Pathogenic were found in two probands (Family 7 and Family 13). Among the multiplex families, identi ed variants did not explain the family aggregation of BCGitis in three of them (Family 2, Family 37 and Family 44) and these families will be subject to further analysis. The proband of Family 2 showed a variant in MSMD-Gene classi ed as Likely Pathogenic ( Figure 1). Among all probands, 70.5% (31/44) presented mild or transient infectious conditions or immunoallergic conditions besides BCGitis. Table 1  In the whole cohort, Pathogenic or Likely Pathogenic variants in IEI-Genes were found in 19/44 (43.2%) of probands. Tables 2 to 4 shows Pathogenic or Likely Pathogenic variants in IEI-Genes, including parental origin of inherited alleles by groups of probands according to clinical phenotype.        Six probands with Pathogenic or Likely Pathogenic variants in IEI-Genes presented clinical phenotype "Only BCGitis" ( Table 2). Among them, four had suppurative lymphadenitis, one associated with injection site abscess and one associated with secondary bacterial infection. Two other probands had regional enlarged lymph node > 3cm. Fifty percent (3/6) presented altered IL12p70-IFNγ axis: one with low IFNγ, two with low IFNγ and low IL-12 and one with low IFNγ and IFNγ receptor not detected. Six variants classi ed as Pathogenic or Likely Pathogenic in six different IEI-Genes were identi ed in WES of those six probands. Among these IEI-genes, 50% (3/6) are related to defects in intrinsic and innate immunity (STAT1, TBK1 and TLR3). The remaining three genes are related to autoin ammatory disorders (NOD2), complement de ciencies (CFH) and immune deregulation diseases (NFAT5).
Three probands with Pathogenic or Likely Pathogenic variants in IEI-Genes presented clinical phenotype "Mild or Transient Symptoms" (Table 3). Two had suppurative lymphadenitis (SL) and one had an ulcer >1 cm (UL). In this group, only 33.1% (1/3) of probands presented alteration in IL12p70-IFNγ axis (low IFNγ) and 66.7% (2/3) probands had transient cytopenia (one neutropenia and one low CD8 T and low NK cells). Three variants classi ed as Pathogenic or Likely Pathogenic in three different IEI-Genes were identi ed in WES of these three probands. One of these variants is compound heterozygosis. These genes are IEI-Gene related to combined immunode ciencies with associated or syndromic features (PMS2), IEI-Gene related to predominantly antibody de ciencies (TNFSF12) and IEI-Gene related to complement de ciencies (CFH).
The largest clinical group of this cohort with detectable Pathogenic or Likely Pathogenic variants in IEI-Genes was composed of 10 probands with clinical phenotype "Immunoallergic conditions" (Table 4). Among them, eight had suppurative lymphadenitis (SL), one associated with injection site abscess (ISA) and another one, associated with secondary bacterial infection (BSI). Two other probands had injection site abscess (ISA). Seven out of 10 (70%) required at least one hospital admission due to asthma or recurrent pneumonia and 8/10 (80%) probands had an altered IL12p70-IFNγ axis (seven with low IL-12, three with also low IFNγ, and one with low IFNγ). Cytopenia was identi ed in 6/10 (60%) of these probands, of which 5/6 (83.3%) were transient cytopenia (neutropenia, total lymphopenia, low CD4 T cells, low naive CD4 T cells, low B cells or low NK cells), and 1/6 (16.7%) had persistent low CD4 T cells. Four out of 10 (40%) probands reported tuberculosis cases in their families and another two probands were familial simplex cases of BCGitis.
WES identi ed 14 variants classi ed as Pathogenic or Likely Pathogenic in 11 IEI-Genes in these 10 probands. Three in 10 (30%) had variants classi ed as Pathogenic or Likely Pathogenic in two or more genes each. Four of the 11 IEI-Genes (36.4%) were related to defects in intrinsic and innate immunity (IFNGR1, IRF8, JAK1 and STAT1), all of them in MSMD-genes. Another 3/11(27.3%) were IEI-genes related to autoin ammatory disorders (CARD14, NLRP12 and NOD2). Another 3/11 (27.3%) were IEI-Genes related to combined immunode ciencies with associated or syndromic features (CHD7, FOXN1, STAT3). One of the 11(9.0%) genes was an IEI-Genes related to predominantly antibody de ciencies (TNFRSF13B). Parental origin of inherited alleles is also shown in Tables 2 to 4. De novo variants classi ed as Pathogenic or Likely Pathogenic were found in 3 IEI-genes: CHD7, STAT3 and one allele of PMS2. One variant in JAK1 was also assumed to be de novo, as it was novel-LOF and it was not found in the WES of the mother and brother (father's sample unavailable for testing). Three other novel-LOF variants were found, two in CHD7 and one in STAT1.  [20]. Moreover, the percentage of IEIs diagnoses based on BCGitis is still uncertain [24].

Discussion
Severe BCG-AE are usually associated with genetic defects or allelic variants associated with IEI affecting innate or adaptive immunity [25]. Although BCGosis is more frequent in patients with severe clinical conditions such as severe combined immunode ciency (SCID), localized BCG-AE have been described in these patients, with incidence rates ranging from 16.6% [12] to one third of patients investigated [26].
In this study, different in silico predictors and a search software, aggregator and impact analysis tool for human genetic variation -VarSome [22] -were used, in observance of ACMG criteria for variant pathogenicity classi cation.
As expected, there was a predominance of IEI-genes related to Defects in Intrinsic and Innate Immunity among variants classi ed as Pathogenic or Likely pathogenic: they represented 35.3% (6/17) of IEI-genes with identi ed variants. Among them, the most frequent were MSMD-genes, with 4/17 (23.5%) classi ed as Pathogenic or Likely Pathogenic variants.
MSMD is a group of rare innate immunity errors characterized by individual selective susceptibility to clinical disease caused by weakly virulent mycobacterial species, such as BCG strains and environmental mycobacteria, in healthy patients with normal resistance to other microorganisms, in the absence of immunological abnormalities evident in routine evaluation [27][28][29].
In countries where BCG is administered early in life, BCG infections are often the rst sign of MSMD disease [28, 30,31], with severity in patients with MSMD varying from localized infections to widespread and life-threatening infections [29,32].
The incomplete penetrance in MSMD genes suggests the presence of modifying factors explaining different susceptibility of individuals with the same variant but with variable clinical outcomes [30]. Allelic heterogeneity can also interfere, as different defects of the same gene can result in different disorders [28,33].
Among variants previously described in the literature and found in this study, one of them occurred in a MSMD-gene: a heterozygous variant in the IFNGR1 gene (Proband 27_1). This variant has been previously identi ed by Jouanguy et al. [34] as a partial IFNGR1 de ciency in one child with curable BCG infection and his sibling with latent tuberculosis. This variant was also reported by Remiszewski et al. [35] who identi ed a 20-year-old female with disseminated Mycobacterium avium disease involving bones, lungs and brain. She was completely healthy until this illness and had been vaccinated with BCG in infancy without complications. Functional analysis of this variant was performed by van de Wetering et al.
[36] and con rmed that the severely reduced function of the I87T mutant receptor can lead to partial IFNGR1 de ciency. IFNGR1 de ciency caused by variants in IFNGR1 gene that are either autosomal recessive or autosomal dominant have a high degree of allelic, cellular and clinical phenotype heterogeneity [37][38][39]. Most recessive IFNGR1 de ciencies result in complete loss of cellular responsiveness to IFNγ due to mutations that preclude the expression of IFNGR1 on the cell surface.
Another group of IFNGR1 de ciencies is due to missense mutations which result in normal expression of IFNGR1 at the cell surface, however, the resulting receptors show either diminished or hindered binding of IFNγ [36]. This is compatible with the alterated IL12p70-IFNγ axis test of proband 27_1.
Siblings reported by Jouanguy et al. [34] had also some degree of atopy, as observed with proband 27_1. Studies have shown that genetic variants in these IFN-pathway genes may have some degree of susceptibility to a range of common, chronic human diseases, which have an in ammatory component with high IgE levels and clinical phenotypes of asthma, atopic dermatitis and eczema herpeticum [40][41][42].
Variants associated to Autoin ammatory Disorders and classi ed as Pathogenic or Likely Pathogenic in IEI-Genes were found in 3/17 (17.6%) of identi ed IEIgenes of this cohort. Interestingly, these variants were observed especially within probands from the group that displayed immunoallergic conditions.
In autoin ammatory disorders, IEI are usually caused by hyperfunction of the immune system, with frequent manifestations of recurrent in ammatory episodes [43]. Since the disease mechanism of various in ammatory disorders is related to increased production of proin ammatory cytokine IL1β and possible deviation from Th1 response to Th17, a less effective containment of BCG replication [44] could justify the BCG-AE presented by these patients.
No proband in the cohort presented a SCID phenotype and no variant classi ed as Pathogenic or Likely Pathogenic in IEI-Genes related to Immunode ciencies affecting cellular and humoral immunity was identi ed. Although BCGitis is reported in SCID patients, BCG dissemination is more likely to occur [6].
Three variants classi ed as Pathogenic or Likely Pathogenic among IEI-genes related to Combined immunode ciencies with associated or syndromic features were identi ed in CHD7 gene. All patients had a phenotype of severe immunoallergic conditions with recurrent infections ( Table 4). Two of the CHD7 variants are novel (probands 14_1 and 32_1) and the other one is a de novo variant (proband 6_1).
Heterozygous pathogenic variants in CHD7 are the most frequent cause of CHARGE Syndrome [45][46][47] and pathogenic CHD7 variant is considered as a major criterion for the diagnosis of CHARGE syndrome (Hale et al., 2016). All three probands with variants in CHD7 will be investigated in depth and will be described in another article.
Proband 14_1 had another variant classi ed as pathogenic among IEI-genes related to Combined immunode ciencies with associated or syndromic features: that was a de novo variant in STAT3. STAT3 acts as a central transcription factor downstream of multiple cytokine and growth factor receptors and thus regulates antimicrobial responses and cell survival [48]. STAT3 mutations can cause autosomal dominate hyper-IgE syndrome (AD-HIES), characterized by elevated IgE levels, persistent eczema, repeated skin abscesses, recurrent pneumonia with abscess and pneumatocele formation, candida infections, peculiar face and skeletal and connective tissue abnormalities [49,50]. This phenotype is compatible with the relevant clinical characteristics presented by proband 14_1, including high IgE levels.
Two of the 17 (11.8%) Pathogenic or Likely Pathogenic variants were related to Predominantly Antibody De ciencies IEI-genes (TNFSF12 and TNFRSF13B). Studies suggest antibodies can also provide protection against intracellular pathogens such as micobacteria by targeting innate immune antimicrobial activity via Fc receptor-mediated opsonization and phagocytosis [51][52][53], which are found in all innate immune cells [54]. Lu et al. [53] demonstrated that different antibody pro les appear to correlate with different stages of TB disease (active or latent) and may be able to lead to cell cytotoxicity mediated by NK cells, phagolysosomal maturation, in ammasome activation or intervene in other defense mechanisms.
The same can occur to humoral immune response to vaccination with BCG [55,56]. In this way, IEI that affect production of antibodies or B cell function may result in ine ciency to contain BCG replication, leading to an adverse event.
One proband from this cohort had a chromosome 17p13 Microdeletion Syndrome (proband 7_1). To date, no IEI-gene was identi ed in this region of chromosome 17 [57][58][59][60]. However, there are two reports associating of 17p13 microdeletion to thymic hypoplasia suggesting partial DiGeorge's Anomaly [61] and one case associated with T cell lymphopenia [62]. However, no monogenic disorder was identi ed in these two studies.
Among the family cases, many had relevant clinical conditions in addition to the BCG-AE, however, the genetic variants found did not meet the criteria adopted in this study for analysis, particularly an explanation for the familial aggregation of BCGitis.
Variants in IEI-genes classi ed as Pathogenic or Likely Pathogenic were found in all groups of clinical phenotypes analyzed ( Figure 2). That is in line with the substantial phenotypic and clinical heterogeneity observed and described within groups of patients with variants in the same gene and even between individuals from same pedigree [63].
Diagnostic sequencing and genetic testing have the drawback that the effect of a variant on function cannot be inferred from sequencing alone, with a large proportion of variants classi ed as variants of uncertain signi cance (VUS) persisting in bioinformatic databases [64].
In this study, many cases with relevant clinical ndings presented VUS variants or even no variants in IEI-Genes were found at all. Ewans et al. [65] observed that even if a disease-causing gene is not identi ed in a rst analysis, or the variants found are initially classi ed as VUS, a future reanalysis increases molecular diagnoses. So, re-analyses will be carried out in due course and other groups of genes will be investigated in the probands of this cohort.
In this study we identi ed several cases in which the proband inherited a candidate causative variant from an unaffected parent. This observation led us to hypothesize a possible incomplete penetrance for those cases. Al Dhaheri et al.
[66] raise four explanations for this sort of observation: a) a misclassi cation of a parent presenting a mild phenotype instead of being unaffected; b) an environmental effect requirement for this speci c genotype; c) a bialelic variant requirement in which the second hit was lost due to technical limitation; and d) an oligogenic disease, in which the patient inherited the second failed gene from the other parent and, due to a technical limitation, it could not be identi ed.
Likewise, it is important to remember that the identi cation of associations between an allele and an observed clinical outcome does not necessarily mean that the allele itself conveys a functional difference [36].  [70].
To identify which variants found in a WES might have functional or neutral changes remains challenging. Indeed, even using a standardized approach, a consensus classi cation is not achieved in 100% of cases even among experts [72]. This study does not intend to offer a de nitive genetic diagnosis, but sought to demonstrate a possible relationship that a localized BCG-AE may indicate an IEI, emphasizing the importance of adequate follow-up and investigation of these patients.
To consider that children with localized BCG-AE may have some genetic-molecular disorder linked to an immunological defect is essential to de ne early intervention and prevent clinical complications and unfavorable outcome usually observed in children with undiagnosed primary immunode ciencies timely.
As far as we know, this is the rst study to assess exome of patients who had BCGitis using a panel of IEI-related genes, with many plausible variants identi ed, suggesting that BCGitis may be signaling an inborn error of immunity.