Introduction

There are over 500 different primary immunodeficiencies (PID) linked to inborn errors of immunity (IEI), 555 of which have been genetically identified to date1. This number continues to increase yearly with advances in genomics and cell biology1,2. The spectrum of PID clinical presentations is wide and varied, based on an increased susceptibility to infections and immune dysregulation (autoimmunity, allergic disorders, and/or malignancy)3,4. This heterogeneity leads to diagnostic difficulty and delay that varies between 9 months and 4.7 years5,6,7,8, with a significant impact on morbidity and mortality5. Several scores and characteristic signs have been established to heighten physicians’ awareness and helping for PID diagnosis9,10,11. Nevertheless, the sensitivity and specificity of these scores is low, and the diagnosis of PID remains a major challenge11,12,13.

There is growing evidence that severe infectious diseases in childhood are attributed to IEI14, even if there is no history of recurrent clinical and biological signs14,15. Gaschignard et al. determined that 16% of invasive pneumococcal infections are related to genetic immunity defects16. Several examples have emerged as models of monogenic susceptibility to infectious disease caused by a limited number of etiopathogenic agents and without any other physical conditions: Herpes simplex encephalitis in patients with variants in the TLR3/IFNRA pathway14,17,18, Mendelian susceptibility to mycobacterial infections with impaired IFN-γ mediated immunity, among others19,20,21.

Recent data reported an incidence of 1 IEI in 2000 births (400 new cases per year), and an estimated prevalence of 9.44/100.000 residents in France9. However, these values are probably underestimated5,22,23,24. Despite the widespread recommendations to screen for PID9, immunological phenotypic and/or genetic assessments after an episode of severe infection leading to hospitalization in intensive care are not carried out systematically, and the term “severe” is imprecise. The prevalence of PID in children presenting with serious infection, regardless of the pathogen, remains for the most part unknown in high-income countries. We aimed to evaluate the feasibility of a systematic PID screening protocol, and the prevalence of PID in a real-life cohort of children admitted for a severe infection in a pediatric intensive care unit (PICU) in France.

Methods

Study design

This retrospective longitudinal study evaluated the feasibility of a PID monitoring protocol after a severe infection. Immunological and/or genetic analysis data were collected for all children included in the present study. The study was approved by the Institutional Review Board (IRB) of Montpellier University Hospital (IRB-MTP_2020_04_202000377), and was registered on ClinicalTrials.gov (NCT04356053). All methods were performed in accordance with the relevant guidelines, regulations and with the principles outlined in the Declaration of Helsinki. According to French Law, this anonymous retrospective study did not require patient signed informed consent, yet, non-opposition from all parents or legal guardians was required and confirmed.

Patient population

All children aged from 1 month to 16 years old, admitted to the PICU at the Montpellier University Hospital due to a severe infection, from January 2018 to December 2020, were included. Inclusion criteria were: admission in pediatric ICU for more than 24 h, severe disease of known infectious etiology (bacterial, viral, or fungal infectious disease), child benefiting from a social security scheme. Exclusion criteria were: prematurity up to 6 months of age, severe infections of unknown etiology, children admitted for isolated respiratory syncytial virus bronchiolitis with no other infection-related complications, previous comorbidity as known primary or secondary immunodeficiency, burn victims, other risk factors for infections as for pneumonia: swallowing disorders, tracheotomy, chronic pulmonary pathology, severe asthma, encephalopathy; for meningitis: cochlear implants, meningeal breach,post-neurosurgical meningitis; for deep infection: implanted material, recent surgery, for cardiovascular decompensation, any other chronic pathology favoring an infection, and death before the follow-up consultation at 3 months.

Data collection and systematic immunological screening procedure

Baseline demographic data were collected for each child. According to the local monitoring protocol, children underwent two follow-up consultations by a specialized physician at 3 and from 12 months post-infection to carry out clinical and biological assessments allowing the detection of possible immunodeficiencies (Supplementary Figure S1).

First consultation at three months post-infection: the three main PID screening scores were evaluated according to medical history prior to the severe infection episode: the ten PID warning signs of the Jeffrey Modell Foundation (JMF)12; the 12 warning signs of the German Patients’ Organization for Primary Immunodeficiencies (DSAI)11; and the CEREDIH’s warning signs (French Reference Center for Hereditary Immunodeficiencies)9 (Supplementary Table S1).

The biological tests performed in routine practice to investigate a potential PID followed French recommendations. Data from complementary immune phenotyping panel, Complement System (CS) assessment and IgG subclasses (Supplementary Figure S1) were collected9. Dihydrorhodamine (DHR) test was not used as a first-line treatment during hospitalization in PICU but was carried out in case of severe pyogenic or fungal infection, at the clinician’s discretion. A proposal to update the vaccination schedule was made during follow-up. Available genetic data consisted of the general primary immunodeficiency panel (by NGS), which tests for 159 diseases25, or panels focused on specific pathologies such as congenital neutropenia or Familial Hemophagocytic Lymphohistiocytosis (FHL). These were performed depending on the clinical and biological signs, according to the clinician and biologist.

Second consultation 12 months after the infection: additional immunological and genetic screenings were analyzed to assess persistent immunological abnormalities, and to explore genetic predispositions to severe infections, if necessary.

An anomaly in the immunological investigation was defined according to laboratory standards and reference ranges for pediatric age groups26,27,28,29. The following were retained as anomalies: cytopenia for full blood count; low levels for immunoglobulins and their subclasses, low levels for complement and post-vaccinal serologies, and high levels of C-reactive protein (CRP) or ferritin. Regarding the T, B, NK immunophenotyping results, the following were retained as abnormal: low levels (number and percentage) of T lymphocytes CD3+, TCD4+, TCD8+, B or NK, and excess of activated T-cells. For the lymphocyte subpopulations, we considered low early naïve and naïve TCD4+, low naïve TCD8+ or increased TEMRA TCD8+, low switched and marginal zone B lymphocyte levels, and increased transitional B level. Anomalies at the first assessment without a formal diagnosis of PID were only retained if still present at the second assessment.

Objectives and outcomes

The primary objective was to evaluate the prevalence of PID in our population. The primary outcome was the rate of patients suffering from PID confirmed following immunological screening according to 2022 guidelines from the International Union of Immunological Societies (IUIS) Expert Committee at 3 months after hospitalization30.

The secondary objective was to assess whether our patients would have been detected as potentially suffering from PID based on the main published screening scores (JMF, DSAI, CEREDIH) before the severe infectious episode. The secondary outcome was the rate of children with a positive screening score for any of the three reference scores at inclusion, to determine the children that would have benefited from prior immune screening.

Statistical analysis

Categorical variables were reported with the number of observations (N) and the frequency (%) of each modality. The 95% confidence intervals (95% CI) were calculated using the exact Binomial method. Continuous variables were reported with median and interquartile range (IQR). All statistical analyses were performed with SAS® 9.4 (SAS Institute, Cary, NC, USA).

Results

Population baseline characteristics

In total, 1125 underage pediatric patients were admitted to the PICU at the Montpellier University Hospital between January 2018 and December 2020. Among these, 54 had a severe infectious disease and met the inclusion criteria, and 46 were analyzed (Fig. 1).

Fig. 1
figure 1

Study flowchart detailing the PID monitoring screening protocol, including immunological and genetic follow-up consultations and screenings performed. GT: Genetic testing; Imm. Immunological; PID primary immunodeficiency.

Full size image

The patients’ median age was 3 years old (IQR 0.92–7.01), and 19 of 46 were girls (41.3%) (Table 1). None of the patients had a previous severe infection with hospitalization in the PICU. Among these patients, 30 (65.22%) had a complete vaccination schedule according to the vaccines recommended in France before 2018 or mandatory after. The meningococcal C vaccine was the most commonly missing, with only 32 (69.56%) of the children vaccinated. None of the unvaccinated children had a meningococcal C infection. Almost all children, 42 (91.3%) had a complete vaccinal scheme for the 13-valent pneumococcal conjugate vaccines (Table 1). None of them had invasive pneumococcal disease due to the serotypes covered by the vaccines.

Table 1 Patients’ baseline characteristics.
Full size table

Diagnosed infections and pathogens

The diagnosed infections were of bacterial origin for 22 patients (47.83%), viral for 18 (39.13%), and fungal for one (2.17%) (Supplementary Table S2). Five cases of co-infection were noted (10.87%). In particular, the coinfections encountered were: rhinovirus/adenovirus/bocavirus; adenovirus/coronavirus OC43/enterovirus; coronavirus NL63/RSV; Escherichia coli/rotavirus; and, Herpes simplex virus 1/Mycoplasma pneumoniae (Supplementary Table S2). The infections were neuro-meningeal for 20 children (43.5%) with meningitis (n = 6, 13.04%), purpura fulminans (n = 5, 10.87%), encephalitis (n = 5, 10.87%), and complex seizure (n = 4, 8.70%) (Fig. 2). The second most frequent infections were ear-nose-throat (ENT) infections in 10 children (21.74%).

Fig. 2
figure 2

Type of infection detected in hospitalized children (n = 46). ENT: ear-nose-throat infection; PIMS: Pediatric Inflammatory Multisystem Syndrome; UTI: Urinary tract infection.

Full size image

Immunological and genetic screening three months after hospitalization

The immunological screening performed three months after hospitalization confirmed the PID diagnosis in three patients, revealing a prevalence of PID of 6.52% in our cohort (95% CI 1.4; 17.9). Among the three children with confirmed genetic PID diagnosis, one was hospitalized for a first episode of bacterial meningitis (Haemophilus influenza b), despite being fully vaccinated. This episode revealed a complete deficiency in Complement protein of C5. Genetic analysis using a targeted NGS panel for PID identified a homozygous 3-nucleotide deletion in the C5 gene (c.960_962del, predicted p.Asn320del). Familial segregation found the same variant but in a heterozygous state in both parents and brother, indicating a heterozygous deficiency in the C5 protein. The second child with a PID diagnosis was evaluated for complex seizure in a viral context, presented a weak vaccine response to pneumococcal 13-valent vaccine and had T CD8 + lymphopenia. Genetic analysis showed a 1409Mb heterozygous large deletion in the chromosomal region 22q11.21 (DiGeorge syndrome). The third PID patient presented with Klebsiella aerogenes septic shock, in the context of an urachal duct abscess, and had severe neutropenia. Genetic analysis confirmed congenital neutropenia, finding a heterozygous deleterious variant c.640G > A (predicted p. Gly214.Arg) in exon 5 of the ELANE gene, which codes for neutrophil elastase (Table 2). The variant was absent in the parents, suggesting that it was de novo.

Table 2 Description of children with confirmed PID at the 3-month evaluation (n = 3).
Full size table

Among the 46 children, 40 (86.95%) had immunological anomalies without a genetic diagnosis of PID. Considering the first line analysis (full blood count; IgA, IgG, and IgM levels; post-vaccinal serology), 25 children (54.35%) presented abnormal results, including two of the three patients with confirmed PID (Table 3). There was a high rate of lymphocyte phenotypic anomalies (57.78%, n = 26), and inflammatory markers (CRP and/or ferritin) were elevated in 11 children (26.83%) three months after the severe infection.

Table 3 Immunological screening at 3 months (n = 46) and 12 months (n = 20).
Full size table

A first genetic assessment was available for nine patients (19.56%), allowing the diagnosis of PID in three cases (33.33%). For the remaining six cases (13.04%), the genetic assessment was performed due to unusual severity or type of infection. Four patients had a PID panel and two a specific panel targeted Familial Hemophagocytic Lymphohistiocytosis. However, these analyses did not identify any PID.

One child underwent a specialized gastroenterology consultation and specific genetic analysis, in the context of an alarming evolution of an acute post-influenza liver failure. This patient was diagnosed with CALFAN syndrome as a result of a SCYL1 deficiency: cholestasis, acute liver failure, and neurodegeneration (Tables 3 and 4).

Table 4 Case description of children with confirmed stable immunological abnormalities at 3 months and 1 year (n = 4) or persistent acute post-influenza liver failure (n = 1), and genetic testing results.
Full size table

Immunological and genetic screening 12 months after hospitalization

One year after hospitalization, 32 of 43 (46.51%) patients with abnormalities at the first investigation had a second follow-up consultation (Fig. 1). The three children with confirmed PID had persistent abnormalities and were followed-up with specific consultations. Among the other patients, 17 underwent immunological screening, and anomalies observed at three months were confirmed in only four cases (23.53%) (Tables 3 and 4). Considering all the cases with immunological evaluation 12 months post-hospitalization, additional immunophenotyping abnormalities were discovered in eight out of 20 children (40%) (Table 3). The most frequent abnormalities one year after severe infection were: low NK lymphocyte levels in five patients (41.18%), and anomaly in the distribution of B cell in five children (25%) (Table 3).

PID prognosis with validated screening scores

The three scores were assessed for all patients based on their medical history before the severe infectious episode, which was therefore not considered in the evaluation. According to the DSAI screening score, four children (8.70%) should have had a previous immune exploration, while none and five (10.87%) should have according to JMF score and CEREDIH criteria, respectively. The positive criteria were family history of PID or similar clinical signs, failure to grow, atopy, and recurrent ENT infections. None of these patients received a definitive diagnosis of PID. Consequently, we observed very poor sensitivity of these three scores in our study (Supplementary Table S3).

Discussion

This study evaluated the feasibility of an extended screening protocol for primary immune deficiencies in children hospitalized in PICU following severe infection, regardless of the pathogen. We identified three cases of genetic disease responsible for PID, with a prevalence of 6.52% (95% CI 1.4; 17.9) in this cohort. These results were consistent with another study conducted at Montpellier University Hospital between 2013 and 2015 that found a prevalence of 8%31. Other studies in the literature focused in invasive bacterial infections, finding a higher prevalence of PID ranging from 10 to 20%16,32,33,34.

In our opinion, systematic screening is essential, as evidenced by one of the PID diagnoses following a status epilepticus triggered by a seemingly benign viral infection. We observed a large proportion of viral infections (39.13%), often overlooked, despite the rise of diseases that specifically predispose people to severe viral infections14. Among patients admitted for severe viral infections, the most frequent viral pathogens were EBV, Influenza A and SARS-CoV-2 (each identified in only 3 patients), meaning the comparative excess of viral infections was not due to the COVID pandemic.

Many studies highlighted genetic susceptibilities to severe infections, and the identification of new genes involved in PID and their phenocopies are constantly increasing15,30,35,36. These studies concerned invasive bacterial infections but also classical models of serious viral infections like HSV encephalitis14, in addition to the more common infections such as severe influenza-related pneumonia37, which led us to conduct this study considering all severe infections regardless of the pathogen.

The present study reinforced the idea that clinical screening scores are insufficient in the event of severe infection, as none of our three PID patients tested positive for JMF, DSAI and CEREDIH scores11,12,13. The low sensitivity found in our work was affected by the low PID prevalence in our population, but also highlights the fact that other criteria are required to diagnose PID in the case of severe infection. These results support both the feasibility and benefits of immunological and genetic screenings for PID post-severe infection.

However, the optimal timing for carrying out the immune assessment remains a complex matter. Consultation and immunological screening at three months of the infectious episode allowed an early detection of the most at-risk immune abnormalities. However, for example, diagnosis of Severe Combined Immunodeficiency (SCID) and combined immunodeficiency (CID) are therapeutic emergencies for infants, and in these cases, control measures are implemented as soon as possible. Ensuring appropriate prophylaxis, vaccinations, specific treatment, and sometimes early detection in siblings are current challenges. On the other hand, three months after severe infection, most patients still presented immunological anomalies (86.95%), which drastically decreased at one year (26.08%), suggesting that immune status was probably still disturbed at three months. This was particularly evident for lymphocyte phenotypes, as we observed abnormalities for 26 children (57.78%) at three months and persisted in only four of 17 cases (23.53%) at one year. Our results suggest that a more in-depth analysis will be easier to interpret if carried out longer than 3 months after the severe infection. Nevertheless, the evolving nature of immunity and variability of immunophenotyping results is a strong argument for an early genetic exploration in the context of immunodeficiency revealed by a severe infection.

Regarding the genetic results, 4 out of 10 patients had positive specific NGS panels, considering also the exploration for liver disease.

Considering a PID prevalence between 6 and 8%, systematic genetic analysis at the onset of severe infection should be considered. A step-by-step approach could be envisaged. An exome analysis, which could be secondarily completed by a trio analysis if necessary (parents and child) would be interesting, as the panels are only looking for a limited number of PID. These results would complement the immunological investigation, for which interpretation could be difficult.

This study presented some limitations. First, this was a monocentric and retrospective study with a small number of patients included. Furthermore, the study took place during the first year of the COVID-19 pandemic, when the rate of infection and severe infections in pediatrics dropped drastically38. This may explain the few PID patients in our study, and it would be interesting to renew it, especially in times of immune debt39. Secondly, the patients hospitalized post-operatively for ENT infections were often hospitalized in intensive care for post-anesthesia reasons, rather than for the severity of their infection. Conversely, severe infections like encephalitis without ICU hospitalization were not included. Moreover, children with Pediatric Inflammatory Multisystem Syndrome (PIMS) were included in the study. However, the pathophysiology of PIMS remains unclear, and the overlap between infection and inflammation/immunology is complex40,41. Deceased patients were excluded, but it would be interesting to study their immunity with post-mortem trio analysis, to propose possible analyses in siblings. It is also important to note the weak vaccine coverage, with only 65.22% of children fully vaccinated. This low vaccination coverage is likely a result of the age of our patients, with 33 (71.74%) born before 2018, start of the French vaccination obligation42, and related to vaccine hesitancy in France43,44. Only 32 children (69.56%) were fully vaccinated for Meningococcus C, and it has been previously shown that missing or incomplete meningococcal C and pneumococcal conjugate vaccines are related to morbidity and mortality of these infections43. The follow-up consultations were important to update the vaccination schedule for these children. We have missing data at both consultations, inherent in any retrospective study. The DHR test was reserved to suggestive bacterial and fungal infections regarding organizational difficulties specific to this test, and screening for detection of autoantibodies to cytokines was not carried out. Missing data from 12-months after hospitalization limited this study by preventing the confirmation of the persistent anomalies and clinical follow-up (recurrence of infections, in particular). Yet, no child was re-hospitalized for a severe infection.

In light of these results, we propose a feasible systematic screening protocol after severe infections comprising full blood count (FBC) analysis and the serum IgG, A, M test (as well as the T, B, NK lymphocyte phenotyping for children under 1 year old to eliminate SCID) during intensive care hospitalization (Fig. 3). We recommend completing the screening at 3 months with post-vaccination serologies (tetanus, pneumococcal), or lymphocyte activation tests if unfeasible (in the case of Ig treatment or too young age), Ig subclasses (in children up to 18 months), T, B, NK phenotyping, and complement analysis in the case of bacterial or fungal infections.

Fig. 3
figure 3

Proposed systematic screening protocol for children admitted in the PICU due to severe infection. Ig: Immunoglobulin, SCID: severe combined immunodeficiency, DHR: dihydrorhodamine, PID: Primary immunodeficiency, FHL: familial hemophagocytic lymphohistiocytosis. 1in children up to 18 months; 2 in children up to 12 months.

Full size image

Specific explorations such as the DHR test following pyogenic or severe fungal infection, or Interferon analysis following a severe viral infection should be performed at any time, depending on the clinical presentation. Finally, the extended phenotyping analysis could be postponed to 1 year, with control of previous abnormal results.

Given the prevalence of PID in this population, and the time needed to perform conventional immunological tests, we also highlight the importance of considering genetic testing at any moment after the severe infection episode. Indeed, these results suggest that severe infections should be contemplated in the national ‘France génomique 2025’ plan, which organizes genome and exome analysis in trio with early sampling. Together with high-throughput techniques, this strategy could facilitate the acquisition of genetic data at an early stage, potentially leading to the identification of new genetic diseases affecting the immune system in this rapidly evolving field.

Conclusion

This study underscored the importance of systematic consultation and immunological screening post-severe infection in pediatric intensive unit patients. The findingsshould raise awareness among pediatricians and physicians caring for children with infections about the need for immunological explorations after severe episodes, regardless of the pathogen, medical history, and age. Early diagnosis of PID is crucialas it may prevent future infections, enable specific prevention and/or treatment, and sometimes early detection in siblings, but remains a challenge. Following classical investigation, a genetic analysis should be readily considered to confirm diagnoses, as well as for research purposes with trio analysis, as the number of genes identified in PID continues to increase.