A multi-composite disease activity score (NFKB1-DAS) for NFKB1 insufficiency Authors: Katharina Thoma, Pia Hassunah, Manfred Fliegauf, Gabriele Müller, Andres Caballero Garcia de Oteyza, Pavla Mrovecova, Helene Kraus, Prerana Agarwal, Maximilian Frederik Russe, Birgit Sawitzki, Lev Petrov, Katrin Vogt, Nadezhda Camacho Ordonez, Aleksandra Hirsch, Christoph B Geier, Máté Krausz, Sigune Goldacker, Natalie Frede, Stephan Ehl, Sarah Salou, Ansgar Schulz, Shahrzad Bakhtiar, Maria Fasshauer, Oliver Hausmann, Anja-Dorothee Hüfner, Martin Stanulla, Anna Luijendijk, Faranaz Atschekzei, Georgios Sogkas, Ulrich Salzer, Baerbel Keller, Michele Proietti, Klaus Warnatz, Bodo Grimbacher Published: 2026 ABSTRACT AbstractNFKB1 encodes the p105/p50 NF-κB1 transcription factor. Heterozygous mutations in NFKB1 may lead to NFKB1 insufficiency affecting multiple organ systems. This study aimed to develop an NFKB1-Disease-Activity-Score (NFKB1-DAS), serving as a tool in future clinical trials. Based on published phenotypic data from 179 patients with 77 distinct NFKB1 variants, we drafted an initial score consisting of 42 parameters. This draft was subsequently analyzed in a multicenter cohort of 41 patients with 22 distinct pathogenic NFKB1 variants. Using a Leave-One-Out (LOO) analysis (cutoff differential value: 23.07), we identified the 21 most relevant parameters to construct the final NFKB1-DAS. Calculation of the NFKB1-DAS in 31 CTLA4-deficient patients as a comparison cohort showed higher scores in NFKB1-insufficient patients (p = 0.03 (95% CI [-0.16, -0.01]) and the NFKB1-DAS correlated with hospitalization (p = 0.01; OR 2.07 (95% CI [1.18; 3.62]). However, given the optimism bias occurring when testing the score in the cohort it was developed with and the limited case numbers, validation of the NFKB1-DAS in an unrelated cohort remains essential. Analysis of the NFKB1-DAS in patients with variants of unknown significance (VUS) in NFKB1 (n = 14) showed individual patients reaching high scores. Future studies could evaluate a possible utility of the score in assessing VUS pathogenicity. Ultimately, the proposed NFKB1-DAS represents a first step toward monitoring disease activity and treatment response in future clinical trials in NFKB1 insufficiency.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1007/s10875-026-02081-3.Keywords: NFKB1, NF-κB1, Disease activity, Treatment, Score, NFKB1-DAS Introduction NFKB1 (Nuclear factor kappa B subunit 1) encodes the cytoplasmic precursor p105, processed to generate the NF-κB transcription factor subunit p50. The NF-κB family comprises the five members NF-κB1 (p105/p50), NF-κB2 (p100/p52), RelA (p65), RelB and c-Rel, being nearly ubiquitously expressed and forming a variety of homo- and heterodimers with activating or repressing functions depending on their combination. The NF-κB signaling system comprises the canonical (NF-κB1-dependent) and the non-canonical (NF-κB2-dependent) pathway. Canonical signaling is typically activated by inflammatory signals (such as TNF-α) via cell surface receptors, which triggers the release of mainly p50:RelA dimers from their cytoplasmic inhibitors (IκB), translocating to the nucleus to regulate their target genes including a variety of immune genes [1–3]. Reflecting extreme intolerance to inactivation, NFKB1 exhibits a pLI of 1.0 and a LOEUF of 0.15, confirming severe constraint against loss-of-function variation [4]. Previous studies demonstrated that most pathogenic heterozygous NFKB1 variants cause (haplo-) insufficiency (HI) of p105 and p50 which may result in a multiorgan disease, referred to here as NFKB1 insufficiency, affecting respiratory, gastrointestinal, lymphatic system and the skin [5]. NFKB1 insufficiency is the most common monogenic cause for primary antibody deficiency [5–7]. However, with 70%, the penetrance of proven pathogenic variants in NFKB1 is incomplete and disease severity varies considerably even among carriers of the same mutation in NFKB1. [5, 8] The disease can progress chronically over many years and remains challenging to treat, resulting in diverse therapeutic strategies. In the cohort reported in this study, the predominant therapeutic approaches were immunoglobulin replacement, immunomodulatory therapy, or antibiotic prophylaxis of infections. Lacking large-scale, systematic treatment studies, the clinical presentation and the experience of the attending physician currently determine the preferred treatment. There is neither a targeted therapy to ‘cure’ NFKB1 insufficiency, nor individualized therapies to manage unique clinical presentations or individual genetic variants [5]. Treatment studies on NFKB1 insufficiency require an objective multi-organ composit score tool to assess the benefits of various therapeutic options. Applying global disease severity scores for inborn errors of immunity, such as common variable immunodeficiency (CVID) severity scores or scores addressing single organ manifestations like the Bhalla score for lung diseases or the NANO scale (Neurologic Assessment in Neuro-Oncology) for neurologic symptoms are not sufficient, since they are not specific enough for NFKB1 insufficiency [9–14]. Our study processes previously known disease manifestations and newly collected data into a novel clinical tool, the NFKB1-DAS. Methods Analysis of previous NFKB1 publications leads to the generation of a preliminary score based on published reports, expert knowledge and review of patients’ records In July 2020 seven main publications were selected describing the NFKB1-associated disease phenotype in 179 patients with 77 distinct NFKB1 variants classified as pathogenic or possibly pathogenic based on prediction tools or functional tests (Table S1). Based on the identified characteristic organ manifestations of NFKB1 insufficiency a first score draft comprising 42 parameters was established (Table S2, Fig. S1 Part 1). Repeatedly reported symptoms and diagnoses were summarized in score parameters, partly using established scores like the CHAI morbidity score (from the ABACHAI study = Abatacept for CTLA4 HAploInsufficiency), the Hartmann score, an immunoglobulin replacement score, the Medical Research Council (MRC) Dyspnoea Scale, the Neurologic Assessment in Neuro-Oncology (NANO) Scale and the Disease Activity Score including C-reactive Protein (DAS-28-CRP) [13, 15–19] or experience from previous studies [22]. The NFKB1-DAS is indicated as a percentage value corresponding to the achieved score points relative to the maximum score value (File S4). The maximum possible score value results from the maximum number of points of all parameters with available data. Zero to three points can be achieved for each parameter (0 = not present, 1 = mild, 2 = moderate, 3 = severe). Score calculation according to the first score draft in a cohort of 41 patients with pathogenic variants in NFKB1 NFKB1 cohort and collection of clinical data From October 2020 to February 2023, prospective and retrospective clinical data collection (going back to 2007) in a cohort of 41 patients with pathogenic variants in NFKB1 was conducted, aiming to calculate the score according to the first score draft (Fig. S1 Part 2). Patients were recruited from clinical databases and eight centers in Germany and Switzerland and have a central European background in the vast majority [20, 21]. Our study cohort comprised 41 patients carrying 22 unique pathogenic NFKB1 variants confirmed by functional in vitro assays to cause a protein defect or predicted to be pathogenic with a high probability (e.g. all truncations) (Tables 1 and 2, Table S3) [5, 6, 8, 23, 24]. In addition, 14 patients with 13 distinct variants of unknown significance (VUS) were collected. Table 1In the study cohort, 35 unique variants in NFKB1 were identifiedNFKB1-VariantIDcDNA changeProtein change(* = stop)AFVariant ‘type’ACMG classificationCADD PHREDFuncionally tested in (PMID)Result of functional testEarly truncation: haploinsufficiency (pathogenic) #1c.159 + 1G > Ap.Ala40Glufs*10 (assumed)NAsplice donor / out-of-frame delEx4Pathogenic/likely pathogenic3326,279,205pathogenic #2c.407 + 1G > Ap.Ile87Leufs*16 (assumed)NAsplice donor / out-of-frame delEx6Pathogenic/likely pathogenic32* #3c.295 C > Tp.Gln99*NAnonsensePathogenic3834,473,196pathogenic #4c.469 C > Tp.Arg157*NAnonsensePathogenic3834,473,196pathogenic #5c.520_521insCTGAp.Leu176*NA4 bp insertion / nonsensePathogenicNA34,473,196pathogenic #6c.584_585delAGp.Glu195Alafs*29NA2 bp deletion / frameshiftNANA* #7c.589delAp.Ile197Serfs*55NA1 bp deletion / frameshiftLikely pathogenicNA* #8c.634_656dupGTGGTGCGGCTCATGTTTACAGCp.Phe220Trpfs*40NA23 bp duplication / frameshiftNANA* #9c.850 C > Tp.Arg284*NAnonsensePathogenic/likely pathogenic4334,473,196pathogenic #10c.872delAp.Asn291Metfs*141NA1 bp deletion / frameshiftPathogenicNA34,473,196pathogenic #11c.904dupTp.Ser302Phefs*71.86e-61 bp duplication / frameshiftPathogenic/likely pathogenicNA34,473,196pathogenic #12c.1066 + 1G > Cp.Phe310Ilefs*76NAsplice donor / out-of-frame delEx11Pathogenic33*pathogenic #13c.1012delTp.Ser338Leufs*94NA1 bp deletion / frameshiftPathogenicNA34,473,196pathogenic #14c.1071_1074delAGAAp.Glu358Lysfs*73NA4 bp deletion / frameshiftPathogenicNA34,473,196pathogenicCentral truncation: heterozygous loss-of precursor (pathogenic) #15c.1361dupAp.Asn454Lysfs*11NA1 bp duplication / frameshiftNANA*assumed to be pathogenic #16c.1365delTp.Val456*NA1 bp deletion / nonsensePathogenicNA32,278,790assumed to be pathogenic by fluorescence microscopy #17c.1819delGp.Ala607Profs*3NA1 bp deletion / frameshiftNANA*assumed to be pathogenicLate truncation (pathogenic) #18c.2670delGp.Ala891Glnfs*6NA1 bp deletion / frameshiftLikely pathogenicNA* #19c.2455 C > Tp.Gln819*NAnonsenseLikely pathogenic****37*Early internal deletion: (p50) haploinsufficiency (pathogenic) #20c.835 + 2T > Gp.Lys244_Asp279delinsAsnNAsplice donor / in-frame delEx9Pathogenic****3326,279,205pathogenicN-terminal half missense: DNA-binding defect (pathogenic) #21c.169 C > Tp.Arg57CysNAmissenseVUS****3334,473,196pathogenicN-terminal half missense: loss-of-expression (pathogenic) #22c.856T > Ap.Tyr286AsnNAmissenseVUS****28834,473,196pathogenicN-terminal half missense: unclear defect (VUS) #23c.150 A > Cp.Gln50HisNAmissenseVUS24341,202,999** #24c.156 A > Cp.Lys52AsnNAmissenseVUS248 #25c.269 A > Cp.Tyr90SerNAmissenseVUS27936,105,815, 34,473,196neutral #26c.307 A > Gp.Asn103Asp6.20e-7missenseVUS****23036,105,815Unknown effect #27c.641G > Ap.Arg214GlnNAmissenseVUS3434,473,196Neutral (unclear in 36105815) #28c.646 A > Gp.Met216ValNAmissenseVUS24334,473,196Neutral (unclear in 36105815) #29c.734 C > Tp.Ala245Val6.20e-7missenseVUS****28234,473,196Neutral (unclear in 36105815) #30c.736 C > Ap.Pro246ThrNAmissenseVUS25534,473,196neutralC-terminal half missense: unclear defect (VUS) #31c.1424 C > Gp.Ala475Gly2.74e-4missenseVUS****486334,473,196Unknown effect*** #32c.1601G > Ap.Arg534His4.73e-4missenseBenign/likely benign29234,473,196Unknown effect*** #33c.1845G > Tp.Leu615Phe2.45e-3missenseBenign/likely benign157234,473,196Unknown effect*** #34c.2548 C > Tp.Arg850Trp1.00e-4missenseVUS3234,473,196Unknown effect*** #35c.2634 A > Tp.Arg878Ser1.36e-5missenseVUS166134,473,196Unknown effect***AF Allele frequency (from gnomAD 01-02-2026); ACMG classification from “ClinVar” (accessed on 2026-07-13) or “https://wintervar.wglab.org” or “https://www.ensembl.org/Multi/Tools/VEP/Ticket?tl=wRgldXpxRXdMh6qh” or PMID 32,278,790, CADD PHRED Combined Annotation Dependent Depletion on PHRED (Phil’s Read Editor) scale, NA Not reported in gnomAD*not functionally tested in vitro but strongly predicted to be pathogenic; **these two variants are neutral/inconspicuous when tested individually but have a pathogenic effect when cis-linked (Fliegauf et al. 2026, PMID 41202999), this was analyzed after data collection of the presented study. The NFKB1-DAS in the carrier was relatively high with 29%; ***reported in PMID 41,202,999 as neutral but performed test not sufficient for evaluating missense variants beyond amino acid 433 (common tests detect p50-specific functions only; p50 comprises the N-terminal half of p105 after cleavage). gnomAD reports high frequencies for variants #31-#34, usually in these cases no pathogenic pathomechanisms are found in functional tests. Variant #32 and #33 are assumed to present polymorphisms (likely benign) due to their high gnomAD frequency; ****low level of confidence of ACMG classification (single submitter or conflicting classifications of pathogenicity); VUS=Variant of unknown/uncertain significance In the study cohort, 35 unique variants in NFKB1 were identified AF Allele frequency (from gnomAD 01-02-2026); ACMG classification from “ClinVar” (accessed on 2026-07-13) or “https://wintervar.wglab.org” or “https://www.ensembl.org/Multi/Tools/VEP/Ticket?tl=wRgldXpxRXdMh6qh” or PMID 32,278,790, CADD PHRED Combined Annotation Dependent Depletion on PHRED (Phil’s Read Editor) scale, NA Not reported in gnomAD *not functionally tested in vitro but strongly predicted to be pathogenic; **these two variants are neutral/inconspicuous when tested individually but have a pathogenic effect when cis-linked (Fliegauf et al. 2026, PMID 41202999), this was analyzed after data collection of the presented study. The NFKB1-DAS in the carrier was relatively high with 29%; ***reported in PMID 41,202,999 as neutral but performed test not sufficient for evaluating missense variants beyond amino acid 433 (common tests detect p50-specific functions only; p50 comprises the N-terminal half of p105 after cleavage). gnomAD reports high frequencies for variants #31-#34, usually in these cases no pathogenic pathomechanisms are found in functional tests. Variant #32 and #33 are assumed to present polymorphisms (likely benign) due to their high gnomAD frequency; ****low level of confidence of ACMG classification (single submitter or conflicting classifications of pathogenicity); VUS=Variant of unknown/uncertain significance Table 2Our study cohort includes 41 patients with pathogenic variants in NFKB1 and 14 patients with a VUS in NFKB1Patients with a pathogenic variant in NFKB1Patients with a VUS in NFKB1Total number of patients4114Male (number/%)16/399/64Female (number/%)25/615/36Mean age at first survey date (years)3842Median age at first survey date (years)4139Minimal age at first survey date (years)29Maximum age at first survey date (years)7570Deceased during the study period (number)20Age at death (years)51 and 54-Cohort characteristics of 41 patients with pathogenic variants in NFKB1, analyzed for the development of the NFKB1-DAS, and 14 patients with VUS in NFKB1. Inclusion criteria were a heterozygous variant in NFKB1 with a frequency in the healthy population of less than 1:100 on gnomAD (if reported) [27]. If a variant was predicted to be “neutral” based on the currently available functional in vitro tests, it was still considered as potentially pathogenic with a frequency of less than 1:100.000 (if reported) and symptoms typically associated with NF-κB1 dysregulation [5, 28]. Without signed consent, deceased, unavailable, with data for < 50% of score parameters or unaffected variant carriers were not included (last criteria chosen to minimize bias in the analysis due to the incomplete penetrance of NFKB1 insufficiency). Carriers were considered unaffected when a maximum of one upper respiratory tract infection (URTI) occurred within the last six months and no other score parameter yielded any points. Cohort characteristics on the CTLA4 comparison group can be found in the supplementary materials Our study cohort includes 41 patients with pathogenic variants in NFKB1 and 14 patients with a VUS in NFKB1 Cohort characteristics of 41 patients with pathogenic variants in NFKB1, analyzed for the development of the NFKB1-DAS, and 14 patients with VUS in NFKB1. Inclusion criteria were a heterozygous variant in NFKB1 with a frequency in the healthy population of less than 1:100 on gnomAD (if reported) [27]. If a variant was predicted to be “neutral” based on the currently available functional in vitro tests, it was still considered as potentially pathogenic with a frequency of less than 1:100.000 (if reported) and symptoms typically associated with NF-κB1 dysregulation [5, 28]. Without signed consent, deceased, unavailable, with data for < 50% of score parameters or unaffected variant carriers were not included (last criteria chosen to minimize bias in the analysis due to the incomplete penetrance of NFKB1 insufficiency). Carriers were considered unaffected when a maximum of one upper respiratory tract infection (URTI) occurred within the last six months and no other score parameter yielded any points. Cohort characteristics on the CTLA4 comparison group can be found in the supplementary materials Functional data for the evaluated NFKB1 variants were synthesized from comprehensive benchmarking studies [6, 23, 25]. The information required for the score was collected based on patient interviews, using questionnaires (File S1 and File S2), and documented clinical data. In general, symptoms that are known to be immune-related were considered as associated with the underlying NFKB1 insufficiency. To calculate the score parameter for programmed cell death protein one positive (PD-1+) T cells, 42 blood samples from 22 patients with pathogenic variants in NFKB1 were analyzed by CyTOF (Fig. S2). For the later following comparison of the NFKB1-DAS in pathogenic vs. VUS patients, PD-1+ T cells were also calculated in 12 patients with VUS in NFKB1 and analysed using Spearman correlation and negative binomial regression (Fig. 3). Patients with pathogenic variants in NFKB1 show an increased level of PD-1+ CD4+ T cells. a There is a positive correlation between the percentage of PD-1+ CD4+ T cells and the NFKB1-DAS. b In patients with pathogenic variants in NFKB1, the percentage of PD-1+ CD4+ T cells is higher than the percentage of PD-1+ CD8+ T cells, which was not seen in patients with VUS in NFKB1 (c). The horizontal line shows median, x shows mean values Score calculation based on the initial score draft Score values were calculated using Excel. One up to seven time points were achieved for each patient. The data used to determine the individual score parameters was not allowed to deviate by more than +/- 3 months from the survey date (resp. -6 months if it was the first survey date). The median time between the first and second time point was 9 months and 6 months between second and third time point. Reducing the number of score parameters using a LOO analysis First, the four CT parameters were integrated into one reducing the total number to 39 parameters. For the following LOO analysis, score data of the first time point from 41 patients with pathogenic NFKB1 variants was used. Only data from the first time point was used to minimize data gaps (not all patients had several assessments) and to avoid potential bias from longitudinal changes. To reduce the number of score parameters, we performed a LOO analysis at the parameter level (Fig. S1 Part 3). For each of the 39 parameters, the overall score was recalculated after omitting that parameter, while keeping the remaining 38 parameters unchanged. The resulting change in score was quantified as a differential value. Specifically, the differential value was defined as the mean squared difference between the original score and the score recalculated without that parameter across all patients. Parameters causing only small changes were considered less relevant and sequentially removed. The remaining parameters with the largest differential values were retained for the final score, reflecting their relative contribution to the overall score. The analysis was performed using R (Fig. 1 and Fig. 2a). The basic principles of this approach are described by e.g. Hastie et al. 2009 (p. 241ff.) or Wang et al., who called the method a Leave-One-Covariate-Out approach, but also included machine learning methods that were not part of the methods in this study [26, 27]. To identify a cutoff to decide which parameters to exclude from the score, parameters were sorted in descending order of their differential value. A cutoff was set between the two parameters with the largest difference in the differential values, as this reflected the largest drop of impact of parameters (Fig. 2). The parameter “other autoimmune/autoinflammatory manifestations” was considered separately since it summarizes multiple parameters, therefore it gains additional data points when other parameters are removed from the score. For instance, when “autoimmune vasculitis” was removed from the score because of a differential value below the cutoff, a vasculitis disease manifestation was scored in the parameter “other autoimmune/autoinflammatory manifestations”. When we repeated the analysis with the 20 parameters above the cut-off only plus “other autoimmune/autoinflammatory manifestations”, we obtained a differential value for the latter parameter in the mid-range, compared to other included parameters (Fig. 2). Therefore “other autoimmune/autoinflammatory manifestations” was included in the final NFKB1-DAS. The remaining 21 parameters form the final NFKB1-DAS presented here (Questionnaire in File S3, Score in File S4). A leave-one-out analysis (LOO) defines a cut-off to identify the 21 most relevant score parameters. A LOO was used to assign a differential value to each parameter (described in Fig. 2a). Parameters were sorted according to this differential value. A cutoff was set between the two parameters with the largest difference in the differential values (see also Fig. 2b) To perform the leave-one-out analysis (LOO), the score was calculated multiple times, each time omitting one of the 42 parameters. a A differential value identified in a LOO analysis defines the effect of each score parameter on the overall score. b A cutoff was set after the biggest differential value difference. c The parameter “autoimmune and autoinflammatory manifestations” shows a differential value in the middle range of all included parameters and was therefore also included in the score Additional analysis with the final NFKB1-DAS investigating its performance Analysis of the NFKB1-DAS in a control group of CTLA4-deficient patients The NFKB1-DAS of 31 affected carriers with pathogenic variants in CTLA4 (cytotoxic T-lymphocyte associated protein 4) (Tables S4, S5 and S7, all variants previously published [28–32]) was compared to the NFKB1-DAS in patients with pathogenic variants in NFKB1 using a two-sided t test to have a first insight whether the score can reflect even subtle, potentially NFKB1-specific differences (Fig. S1 Part 4) [33]. The CTLA4 cohort matched our NFKB1 cohort very well regarding age and sex. Analysis of hospitalization and correlation with the NFKB1-DAS We chose hospitalization as a general marker to test the ability of the NFKB1-DAS to measure disease activity as a first proof-of-concept approach. NFKB1-DAS values of 41 patients with pathogenic variants in NFKB1 were correlated with the number of days spent in hospital within the last 6 months using a hurdle model with negative binomial regression to account for overdispersion in our limited dataset (Fig. S1 Part 4). In all cases, the indication for hospitalization was classified as probably related to the underlying NFKB1 insufficiency. Information on hospitalization was taken from medical records and patient questionnaires. Calculation of the NFKB1-DAS in patients with VUS in NFKB1 During patient recruiting we found 14 patients with VUS in NFKB1 (Table S3) and decided to calculate the NFKB1-DAS in these patients to examine differences in score values over time to see whether patients with pathogenic variants would show a more severe course of disease according to the NFKB1-DAS (Fig. S1 Part 4). Patients with VUS in NFKB1 were examined and scored using the NFKB1-DAS as described before with a median time between first and second appointment of 6 months. The Online Repository provides details on liver manifestations, comprehensive details on the publications analyzed for the development of the initial NFKB1-DAS draft, as well as the initial draft itself. A complete list of all patients included in the study—carriers of either NFKB1 or CTLA4 variants—is presented, alongside the specific CTLA4 variants identified within the cohort. Table S6 summarizes the performance of the NFKB1 patient cohort in the final NFKB1-DAS. Additionally, the questionnaires used for data collection in this study are included. File S3 (questionnaire) and File S4 (calculation procedure) can be used for future calculation of the NFKB1-DAS Results Data from a heterogenic NFKB1 cohort with wide age distribution leads to the NFKB1-DAS comprising 21 score parameters After performing the LOO analysis with data from our multicenter cohort, we present the NFKB1-DAS (File S4) consisting of the following parameters: Frequency of respiratory infections. Frequency of respiratory infections. frequency of antibiotic treatments (for respiratory infections). frequency of antibiotic treatments (for respiratory infections). frequency of non-respiratory infections. frequency of non-respiratory infections. severity of non-respiratory infections. severity of non-respiratory infections. MRC dyspnoea Scale. MRC dyspnoea Scale. Stool frequency/quality. Stool frequency/quality. Gastrointestinal involvement. Gastrointestinal involvement. Severity of skin involvement. Severity of skin involvement. percentage (%) of skin affected by disease. percentage (%) of skin affected by disease. Number of other autoimmune/autoinflammatory manifestations. Number of other autoimmune/autoinflammatory manifestations. Liver disease. Liver disease. Spleen size. Spleen size. Soluble Interleukin-2 (sIL2) receptor level. Soluble Interleukin-2 (sIL2) receptor level. Immunoglobulin G (IgG) level. Immunoglobulin G (IgG) level. Immunoglobulin A (IgA) level. Immunoglobulin A (IgA) level. Immunoglobulin M (IgM) level. Immunoglobulin M (IgM) level. Lymphocytes. Lymphocytes. CD4 + T cells. CD4 + T cells. percentage (%) of HLA-DR+ CD8 + cells. percentage (%) of HLA-DR+ CD8 + cells. percentage (%) of PD-1 + CD4+ cells. percentage (%) of PD-1 + CD4+ cells. percentage (%) of switched memory B cells. percentage (%) of switched memory B cells. NFKB1-DAS in cohort of patients with pathogenic variants in NFKB1 Most patients with pathogenic variants in NFKB1 showed hypogammaglobulinemia (81%) and splenomegaly (76%). Respiratory infections affected more than half of patients (60%). 40% of patients had skin manifestations (especially rosacea, psoriasis and vitiligo), but only 8% with systemic treatment. One third of patients (36%) suffered from gastrointestinal manifestations (36%) or liver involvement (32%) (Table S6, additional information on liver manifestations can also be found in the supplemental data). Analysis of the PD-1 expression of CD4 + and CD8 + T cells in NFKB1 insufficient patients The PD-1 expression of CD4 + T cells was increased in 57% of patients. A Spearman correlation revealed an association between the NFKB1-DAS and CD4+ PD-1+ T cell levels in patients with pathogenic NFKB1 variants (ρ = 0.73, p < 0.001; Fig. 3A). Consistent with this finding, a negative binomial regression demonstrated a 1.37-fold increase of the CD4 + PD-1+ T cell counts per 10-percent-point increase of the NFKB1-DAS (incidence rate ratio (IRR) 1.37; p < 0.001 (95% CI [1.18;1.60]). They show a higher percentage of PD-1+ CD4+ T cells (mean/median: 38%/31%) than PD-1+ CD8+ T cells (mean/median: 18%/15%) in patients with pathogenic variants in NFKB1 (Fig. 3B), which we did not find in patients with VUS in NFKB1 (mean/median of PD-1+ CD4+ T cells 19%/18%, mean/median of PD-1+ CD8+ T cells 18%/13%) (Fig. 3C). Autoimmune manifestations are affecting a variety of organs and appear to disproportionately affect women 54% of patients with pathogenic variants in NFKB1 had at least one autoimmune or autoinflammatory disease not scored in other parameters. Female patients were affected more frequently. Manifestations included the gastrointestinal system, skin/hair, blood vessels, joints, liver, blood cells and lungs (Fig. 4). NFKB1 patients show a variety of autoimmune and autoinflammatory manifestations, with a higher incidence in women. Female patients with pathogenic variants in NFKB1 show a higher frequency of autoimmune and autoinflammatory manifestations, both in a lifetime and b current data (= at the study date). ”Other” in “Autoinflammation ever”: Bone marrow pathology associated with immunodeficiency. “Other” in “Autoimmunity ever”: Autoimmune encephalitis Additional analyses investigating the performance of the NFKB1-DAS The NFKB1-DAS correlates with the hospitalization rate We performed a correlation of the NFKB1-DAS with the number of days spent in hospital in the last 6 months (Fig. S1 part 4). Patients were stratified into a hospitalized group (≥ 1 hospital day within the last 6 months) and a non-hospitalized group. A hurdle model showed that every 10-percent-point increase in the NFKB1-DAS score was associated with a doubling of the probability of hospitalization, defined as at least one day spent in hospital in the last 6 months (p = 0.01, OR 2.07, 95% CI [1.18; 3.62], mean/median NFKB1-DAS hospitalized group: 53%/58%, mean/median NFKB1-DAS non-hospitalized group: 28%/27%) (Fig. 5a). Based on a descriptive analysis of the hospitalization rate, we propose classifying NFKB1-DAS values into mild, moderate, and severe categories (Fig. 6): Mild defined as an NFKB1-DAS < 10% (those patients were not hospitalized in our cohort), moderate defined as an NFKB1-DAS < 30% (with a hospitalization rate of < 10% in our cohort), and severe defined as an NFKB1-DAS of 30% or higher (with a hospitalization rate of > 10% in our cohort). The NFKB1-DAS correlates with hospitalization and is higher in NFKB1insufficient patients than in patients with CTLA4 deficiency. a The NFKB1-DAS correlates with hospitalization defined as at least one day spent in hospital in the last 6 months. b Patients with NFKB1 insufficiency show higher NFKB1-DAS values than patients with CTLA4 deficiency. The horizontal line shows median, x shows mean values Based on the hospitalization rate, the NFKB1-DAS is classified in mild, moderate and severe categories. NFKB1-DAS values of our cohort of patients with pathogenic variants in NFKB1 and the respective hospitalization rate led to a proposed classification of mild, moderate and severe NFKB1-DAS values Patients with pathogenic variants in NFKB1 show higher NFKB1-DAS values than patients with variants in CTLA4 For a first approach testing whether the NFKB1-DAS differentiates between NFKB1 insufficiency and other familiar phenotypes with incomplete penetrance, the scoring system was applied to a cohort of 31 patients with CTLA4-insufficiency as a control group (Fig. S1 Part 4, Table S4, S5, S7). CTLA4-deficient patients reached a mean NFKB1-DAS of 23% (median 22%) compared to 32% in the NFKB1 cohort (median 28%) (Fig. 5b), which was significant with p = 0.03 (95% CI [-0.16, -0.01]). Patients with pathogenic variants in NFKB1 show an increasing NFKB1-DAS over time compared to patients with VUS in NFKB1 In patients with a pathogenic variant in NFKB1, mean and median NFKB1-DAS increased over time (Fig. 7a). In contrast, patients with a VUS in NFKB1 had decreasing NFKB1-DAS scores over time (Fig. 7b). Longitudinal analysis of individual score parameters reveals divergent trends: Some parameters increase (e.g., spleen size or the percentage of PD-1⁺CD4⁺ T cells), while others decrease (e.g., severity of non-respiratory infections) or remain stable (Fig. 7c). These findings suggest that certain parameters may be more informative for tracking disease progression than others. An analysis of the NFKB1-DAS over several study dates may indicate different disease progressions. a and b Patients with pathogenic variants in NFKB1 show an increasing NFKB1-DAS over time (A), whereas patients with VUS in NFKB1 show a decreasing NFKB1-DAS (B). (n = No. of patients). c Individual score parameters show divergent trends over time. In brackets: No. of calculations, one calculation = comparison of data from two study dates of a patient. The horizontal line shows median, x shows mean values Almost all patients with a pathogenic variant in NFKB1, as well as patients with a VUS in NFKB1 received systemic drug therapy (85% and 86% respectively). Children and adolescents < 18 years of age had a lower median NFKB1-DAS value (26%) than patients > 18 years of age (32%). Discussion Selecting score parameters Following our literature analysis in 2020, a large meta-analysis has been published in 2024 by Fathi et al. [34] The results of this meta-analysis corresponded well with the parameters we selected for our NFKB1-DAS. The only manifestations not covered in the NFKB1-DAS are cardiovascular, neurological, and malignant manifestations, which were reported by Fathi et al. with a frequency of 0.6%, 1.7% and 5.6%, respectively. On purpose, we did not include malignancies in the NFKB1-DAS to avoid bias due to chemotherapy. Likewise, the IDDA (Immune Deficiency and Dysregulation Activity) 2.1 score also excludes malignancies from scoring [14]. Recruiting patients Since we recruited patients from immunology departments, there could be a bias regarding disease manifestations that are treated in other departments such as pyoderma. As the clinical picture of NFKB1 insufficiency is primarily known in the context of immune dysregulation and therefore among immunologists, interdisciplinary collaboration and raising awareness of NFKB1 insufficiency in other disciplines is essential to identify affected variant carriers. Data collection NFKB1 insufficiency is a rare disease which limits the number of patients in our study cohort. Data was collected during routine appointments which were often cancelled or postponed during the Covid-19 pandemic, both factors limiting a complete, systematic and broad data collection. We met these challenges by using a multicenter approach and extending our observation period to 2.5 years of prospective, as well as additional retrospective data collection. However, all presented analyses should be carefully interpreted considering the limited data. Systematic cohort analysis complements previous findings on the NFKB1 phenotype In our study we were able to perform a systematic extensive multicenter data collection on 41 patients with pathogenic variants in NFKB1 as well as 14 patients with a VUS in NFKB1 including data that have - to our knowledge - never been assessed before in NFKB1 patients, such as PD-1+ T cell analysis. All variants have been reported before, but not all patients have been published with their full clinical picture. With the exception of a higher frequency of liver manifestations, our phenotypic data assessment fits the findings published in the largest NFKB1 cohort to date by Lorenzini et al. [5] The overall higher proportion of women in our cohort (61%, Table 2) does not fully explain the higher proportion of women with autoimmune manifestations (median 71%; Fig. 4). Higher estrogen levels or incomplete X-chromosome inactivation could be contributing factors [35–38]. Development of final NFKB1-DAS The LOO approach allowed us to use all available patient data effectively, provided a transparent view of each parameter’s impact, and ultimately resulted in a more concise and interpretable score, even when starting from many parameters and a relatively small dataset. However, to mitigate the risk of overfitting inherent to this study design, these findings require validation in independent, heterogenous and ideally larger cohorts. Rare but severe disease manifestations may be underrepresented, as the LOO method emphasizes more frequent manifestations. At the same time, this focus ensures that the score reflects the typical and most observed phenotype and is feasible in clinical practice (one-page questionnaire in File S3). The cohort analysis combined with the LOO approach brings a significant improvement of methodology e.g. compared to the CHAI Morbidity score, developed solely based on expert knowledge [17]. Since the NFKB1-DAS can be calculated for the patient’s current disease state or cumulatively for the entire medical history, it can be used for prospective or retrospective studies. By calculating a score value in percent, we address the problem of missing data in clinical application. All parameters can be collected during regular clinical patient care. Also the PD-1+ CD4+ T cell analysis can be readily implemented in routine clinical laboratories (at a cost of approximately 15€) using a simple four-color whole-blood flow cytometry panel (CD45, CD3, CD4, and PD-1) on standard diagnostic instruments, making its translation into clinical practice technically feasible and relatively inexpensive. The NFKB1-DAS may uncover pathogenic effects of NFKB1 variants classified as VUS For the VUS identified in this study, no mechanistic insights into the pathomechanism of the resulting protein can be provided. This is a common limitation for VUS, where not only the clinical significance but often also the primary consequences at the protein function level remain unresolved. At the same time, ’neutral’ assessments in the currently available functional tests (which do not cover all protein functions) do not necessarily confirm a variant to be benign. On the contrary, particularly rare variants with strong variant effect predictions (VEPs), which were indistinguishable from controls in these assays, cannot be considered as necessarily non-pathogenic. We hypothesize that a high NFKB1-DAS value in patients carrying a VUS in NFKB1 indicates NF-κB1 dysregulation due to a potentially pathogenic effect of the VUS. For instance, patients NFKB1-16 (carrying NFKB1-variant-30, c.736 C > A, p.Pro246Thr), NFKB1-20 (carrying NFKB1-variant-24, c.734 C > T, p.Ala245Val), and NFKB1-23 (carrying NFKB1-variant-29, c.269 A > C, p.Tyr90Ser), who all carry neutrally tested variants, have high score values of 36.84%, 35.71% and 42.11% (classified as severe according to our proposed scheme demonstrated in Fig. 6) at the first time point, respectively [6, 23]. Therefore, it could be worthwhile re-analyzing these variants for a functional defect as a first approach to test our hypothesis, followed by studying a possible correlation between the NFKB1-DAS and the pathogenicity of NFKB1 variants in unrelated and larger cohorts. Moreover, the contribution of concomitant secondary e.g. (epi)-genetic factors should be considered. One example is patient NFKB1-37, who carries a VUS in NFKB1 (NFKB1-variant-25, c.307 A > G, p.Asn103Asp) and a variant in TNFRSF13B (tumor necrosis factor receptor superfamily number 13B), encoding TACI (Transmembrane activator and CAML interactor) (c.579 C > A, p.Cys193*) who has a high NFKB1-DAS of 24%. Performance of the NFKB1-DAS The NFKB1-DAS was higher in NFKB1 insufficient patients than in patients with CTLA4 deficiency, although CTLA4-deficient patients reached a mean NFKB1-DAS of 23%, which is due to shared phenotypic characteristics, such as hypogammaglobulinemia, autoimmune, respiratory and gastrointestinal manifestations [5, 33]. Our multi system composite NFKB1-DAS aims to offer a more specific representation of NFKB1 insufficiency than previous scores developed for different or multiple diseases, such as the CHAI morbidity score or the IDDA 2.1 score, and to be a more suitable tool for treatment studies that rely on depicting the effects on specific disease manifestations as accurately as possible [14, 17]. The presented performance analyses (comparison with CTLA4 cohort and hospitalization analysis) should be interpreted cautiously and perceived as proof-of-concept approaches, considering the limited sample size resulting in high variance and deviation from standard distribution. These analyses were also not adjusted for age, sex, or disease duration due to the limited data available. Conclusions regarding sensitivity and specificity of the NFKB1-DAS should not be drawn, instead, validation of the NFKB1-DAS should be performed in a follow-up study unrelated to the study cohort the score development was based on, to minimize bias. The proposed severity levels (Fig. 6), should be examined and further developed in these follow-up studies. Application of the NFKB1-DAS to guide medication strategies and track treatment success The score data of this cohort demonstrates that disease activity increases with age even under medical treatment. This fits the long-assumed age-dependency of NFKB1-insufficiency and indicates that a disease-specific treatment is still missing. Follow-up studies should test this finding, since the number of patients with multiple score analyses was limited. NFKB1 insufficiency can lead to hyperinflammatory disease similar to A20 haploinsufficiency, where a negative regulator of NF-κB is reduced and NF-κB signaling is overactive [39]. Whether anti-inflammatory drugs like TNF-α inhibitors are superior to other treatment options beyond individual cases is currently investigated in a worldwide NFKB1 study using and validating the NFKB1-DAS. An important limitation in the assessment of complex chronic diseases, one that is also reflected in our score, is the unclear distinction between acute disease activity and chronic, permanent organ damage. Follow-up studies may be able to further refine the NFKB1-DAS in this regard. Conclusion Increasing NFKB1-DAS values in affected NFKB1 insufficient patients underline the urgent need for more efficient personalized treatment options. Our results pave the way for using the NFKB1-DAS as an objective tool for treatment studies in patients with NFKB1-insufficiency. Future studies should assess the ability of disease activity monitoring and potential diagnostic value of the proposed composite score. Supplementary Information Below is the link to the electronic supplementary material. Supplementary File (PDF 2.12 MB) Acknowledgements We thank all patients and their families for participating in this study. We thank all colleagues who made our extensive data collection possible: University Clinic Freiburg (Germany): Monika Erler, Kerstin Mörke, Anja Lübbers, Tanja Mattmüller, Juli-Christin von Guilleaume and Henrike Ritterbusch; Helios Clinic Krefeld (Germany): Nina Brauer, Sandra Goertz and Beate Lamers; St. Georg Clinic Leipzig (Germany): Anna-Hedwig Müller; University Clinic Frankfurt (Germany): Gudrun Sach; Department of Pediatrics, University Medical Center Ulm (Germany): Sandra Steinmann; Department for Pediatric Rheumatology and Immunology, St. Georg Clinic, ImmunDefektCentrum Leipzig (IDCL) (Germany): Michael Borte. We also thank Marlon Grodd and Claudia Giuliani from the Institute of Medical Biometry and Statistic of the Albert-Ludwigs-University of Freiburg (Germany) for their support regarding statistical analysis and the Genetics and Genomics Unit (GGU) and the Advanced Diagnostics Unit at the CCI Freiburg (Germany) for their support regarding genetic diagnosis and immunophenotyping. We thank the Biobank of the CCI for their support with collecting and storing patient blood samples. We thank the Lighthouse Core Facility of the University Clinic Freiburg (Germany) for providing access to essential technologies for our analysis. Thanks to Hannah-Philine Dey for her work with the PD-1 data analysis. Thanks to Stephan Rusch and Pauline Ines Staus for helping with the ESID/GAIN Registry. Thanks to Ulrich Baumann, David Friedmann and Jens Thiel for giving feedback on Score-Parameters. Thanks to Teresa Sprang for her organizational help. We highly appreciate all contributions to our ongoing world-wide NFKB1 cohort study for monitoring treatment success in NFKB1-disease. Questionnaires used in these studies are provided in the Online Repository or upon request and can be sent to bodo.grimbacher@uniklinik-freiburg.de. We thank the funding organizations of this project: This project was funded by internal funds of the University Clinic Freiburg and the Albert-Ludwigs University Freiburg. B.G. is funded by the Deutsche Forschungsgemeinschaft (RESIST – EXC 2155 – Project ID 390874280; CIBSS – EXC-2189 – Project ID 390939984; SFB1160/3_B5; and GR 1617/19-1 – project #555820492, GR 1617/17-1 – project #519635399); the EU-funded PhD program IMMERGE (https://immergeproject.eu); the BMBF rare disease program (GAIN 01GM2206A); and the Wilhelm Sander-Stiftung, Förderantrags-Nr.2023.115.1. Samples were obtained with help of the CCI-biobank, a partner of the FREEZE Biobank Freiburg. We acknowledge the support by the Open Access Publication Fund of the University of Freiburg. Authors’ contributions Conceptualization: Bodo Grimbacher, Klaus Warnatz. Methodology : Katharina Thoma, *Genetics* : Manfred Fliegauf, Michele Proietti, *Statistics* : Katharina Thoma, *first NFKB1-DAS draft* : Bodo Grimbacher, Klaus Warnatz. Software for CT scan analysis : Maximilian Frederik Russe. Formal analysis : Katharina Thoma, *Genetics* : Manfred Fliegauf, Michele Proietti, Andres Caballero Garcia de Oteyza and Pavla Mrovecova. Investigation : *blood samples* : Helene Kraus and Baerbel Keller, *CT scan analysis* : Prerana Agarwal, *PD-1 data* : Birgit Sawitzki, Lev Petrov, and Katrin Vogt, *patient data* : Katharina Thoma, Bodo Grimbacher, Pia Hassunah, Gabriele Müller, Nadezhda Camacho Ordonez, Aleksandra Hirsch, Christoph Geier, Máté Krausz, Sigune Goldacker, Natalie Frede, Ulrich Salzer, Stephan Ehl, Sarah Salou, Ansgar Schulz, Shahrzad Bakhtiar, Maria Fasshauer, Oliver Hausmann, Anja-Dorothee Hüfner, Martin Stanulla, Anna Partheil, Faranaz Atschekzei, and Georgios Sogkas. Resources : Bodo Grimbacher, *PD-1 data* : Birgit Sawitzki. Data Curation : Katharina Thoma. Writing - Original Draft : Katharina Thoma. Writing - Review & Editing : Manfred Fliegauf, Bodo Grimbacher, Katharina Thoma. Visualization : Katharina Thoma. Supervision and project administration : Bodo Grimbacher. Funding This project was funded by internal funds of the University Clinic Freiburg and the Albert-Ludwigs University Freiburg. B.G. is funded by the Deutsche Forschungsgemeinschaft (RESIST – EXC 2155 – Project ID 390874280; CIBSS – EXC-2189 – Project ID 390939984; SFB1160/3_B5; and GR 1617/19 − 1 – project #555820492, GR 1617/17 − 1 – project #519635399); the EU-funded PhD program IMMERGE (https://immergeproject.eu); the BMBF rare disease program (GAIN 01GM2206A); and the Wilhelm Sander-Stiftung, Förderantrags-Nr.2023.115.1. Samples were obtained with help of the CCI-biobank, a partner of the FREEZE Biobank Freiburg. We acknowledge the support by the Open Access Publication Fund of the University of Freiburg. Data availability The primary data are available from the corresponding author upon reasonable request. Declarations Disclosure of conflicts of interest The authors declare no conflicts of interest with regard to this work. Ethics This study was conducted in accordance with the recommendations for studies with human subjects and the declaration of Helsinki and approved by the Ethics Committee of the Albert-Ludwigs-Universität Freiburg (Nr 93/18 220302 last amended on 2022/12/22). Informed consent was obtained from all study participants or their legal guardian in case of minors. All data was anonymized to ensure participant confidentiality. No financial benefits were gained from the conduct of this study by either the investigators or the participants. Competing interests The authors declare no competing interests. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Katharina Thoma, Email: Katharina.thoma@uniklinik-freiburg.de. Bodo Grimbacher, Email: bodo.grimbacher@uniklinik-freiburg.de. References Associated Data Supplementary Materials Supplementary File (PDF 2.12 MB) Data Availability Statement The primary data are available from the corresponding author upon reasonable request.