Familial Mediterranean fever (FMF) is a genetic autoinflammatory disease characterized by mutations within the MEFV (MEditerranean FeVer) gene that encodes the pyrin protein (French FMF Consortium, 1997; The International FMF Consortium, 1997; Papin et al., 2000).

Despite being considered a rare disease worldwide with an estimate of over 200,000 affected people (Gallego et al., 2023), it is common in populations of Mediterranean descent, including Armenians, Turks, Jews, Greeks, and Italians with a prevalence ranging from 10 to 1,500 per 100,000 population (Cerrito et al., 2015; Alparslan et al., 2020). Moreover, FMF prevalence (13–20/per 100,000) is notable in Japan (Migita et al., 2016), and recently it has also been reported in China (Wu et al., 2018).

As of today, 399 variants in the MEFV gene have been identified and their number is increasing with the development of sequencing approaches (Tufan and Lachmann, 2020). The majority of 63 known pathogenic/likely pathogenic variants are located on exon 10 which encodes the B30.2/SPRY domain, responsible for the activation of caspase-1 (source: Infevers database, https://infevers.umai-montpellier.fr/web/search.php, last accessed 15 August 2024). The most common variant of MEFV is M694V (c.2080A > G), which is also a dominant variant in the Armenian population (Hayrapetyan et al., 2016). Other frequent exon 10 variants are M694I (36.0%), V726A (29.5%), M680I (26.2%), R761H (4.9%). These variants constitute almost 75% of all FMF patients and are associated with more severe disease courses among Mediterranean populations. Up to two-thirds of registered variants distributed over the entire gene are either not classified or classified as variants of uncertain significance (VUS) due to their unknown clinical association (Accetturo et al., 2020).

FMF is primarily an autosomal recessive disorder, however, this pattern is not uniform across all cases, adding complexity to genetic counseling and risk assessment (Manukyan and Aminov, 2016). Many individuals with FMF are compound heterozygotes. Different combinations of mutations exhibit a wide range of FMF phenotypes diverse in their symptoms and intensity, indicating incomplete penetrance (Procopio et al., 2018).

Clinical and research laboratories in Armenia use PCR/qPCR-based approaches to test the presence of the most prevalent 12–26 pathogenic mutations, such as FMF StripAssay (12 mutations, ViennaLabs, Austria) or FMF Multiplex real-time PCR kit (26 mutations SNP Biotechnology, Turkey). The common disadvantages of these kits are the limited scope of mutations tested (12 or 26 of known 63), risks of interference for closely located mutations (for example, M680I (G/C-A), I692del, M694V, K695R) in available qPCR tests, and limited capacity for further multiplexing. These limitations can be overcome with Sanger and next-generation sequencing (NGS). Sanger sequencing is considered a gold standard for accuracy and offers opportunities for studying genetic variability on the scale of an exon or a full gene (Schmidt et al., 2022). However, it has been proven that the costs of Sanger sequencing per sample are very high due to the small target range (∼1,000 nucleotides in a single run) limiting its use in clinical settings (at least in low- and middle-income countries (LMICs), such as Armenia). Furthermore, sequencing a single gene like MEFV using high-throughput Illumina NGS platforms is both technically challenging and costly. The installation and operational costs of these instruments require significant investments in infrastructure and a large number of samples is required per run to make it cost-effective. In contrast, nanopore sequencing (NS) developed by Oxford Nanopore Technologies (ONT) offers a more cost-effective solution, with capital and consumable costs as low as approximately $1,000, making it particularly suitable for use in settings with limited laboratory infrastructure. NS is actively explored for the identification of pathogenic variants in various genetic disorders, such as Familial Hypercholesterolemia (Soufi et al., 2022), tandem repeat disorders (Stevanovski et al., 2022), Werner Syndrome (Miller et al., 2021), sickle cell disease (Christopher et al., 2021), and others (Minervini et al., 2020; Oehler et al., 2023).

In this study, we focused on developing an affordable nanopore-based full-gene sequencing test for detecting MEFV mutations to assist in the diagnosis of FMF.

A total of 14 patients (5 males and 9 females) and 2 healthy controls (2 females) underwent full MEFV gene nanopore sequencing (NS). The DNA samples were sequenced with library preparation using R9 or R10 chemistry on both Flow cells and Flongle flow cells. Additionally, whole genome sequencing data generated on the Illumina platform was available for some of the samples (see Table 1).

Nanopore sequencing statistics are shown in Figure 1. Using five tiling fragments, we achieved near-complete coverage of the full-length MEFV gene with balanced sequencing depth overall amplicons. The detailed statistics for individual runs are provided in Supplementary Data Sheet S1. The comparison of variants detected by NS and qPCR demonstrated a high level of concordance (Table 2). We performed a performance analysis of NS variant calling using qPCR results as the ground truth. Compared with qPCR the balanced accuracy of NS detection was 0.97 with a sensitivity of 0.93, specificity of 1.00, and F1 score of 0.97. From three NS replicates, two showed complete concordance of detected variants (samples F5 and F8). In one replicate of the F13 sample M694V was detected in a heterogeneous state (with V10 chemistry and R9.4 flow cell), while in the other replicate (with V14 chemistry and R10.4 flongle flow cell), the M694V variant was detected in a homozygous state concordant with qPCR results. Furthermore, in another sample (F19) additional pathogenic mutation V726A in a heterogeneous state was detected. Unfortunately, we did not have WES data for this sample to confirm the NS or qPCR results (Table 2).

Moreover, both NS and WES showed additional variants outside the scope of the PCR mutation panel (Supplementary Data Sheet S2). For example, both WES and NS identified the R202Q (Supplementary Figure S2). In addition, 19 intronic, 5′- and 3′- UTR variants were discovered with NS (Supplementary Data Sheet S2).

We present a protocol for full MEFV gene sequencing with accuracy comparable to qPCR for detecting common variants. Unlike qPCR, our method is not restricted to a predefined set of variants or locations and can discover additional clinically relevant variants that qPCR will miss. In contrast to WES, it enables the detection of variants not only in exons but also in intronic and regulatory regions. Although all variants observed in noncoding variants in the 3′UTR and intronic regions were previously annotated in ClinVar and classified as benign (see Supplementary Data Sheet S2) this also shows a possibility to identify pathogenic mutations located in non coding regions of the gene such as rs773992396 G>C located in intron 6 (Balta et al., 2020). Furthermore, the sequencing protocol enables more in-depth studies to elucidate the role of coding variants in disease. For instance, while the R202Q variant is classified as benign, recent studies have highlighted its potential association with an inflammatory FMF phenotype (Kandur et al., 2022). Our protocol has also proven effective with new sequencing chemistry and Flongle flow cells. The latter, especially when used with 12-sample multiplexing, makes full-length MEFV gene sequencing both cost- and time-efficient, with results obtainable within 12–24 h including bioinformatics analysis and report generation. Additionally, tree samples were analyzed in replicates, and the variant calling results were consistent across different nanopore sequencing chemistries, aligning with the qPCR genotyping results. In one sample (F13), the correct genotype was detected using the new V14 but not the V10 chemistry suggesting that switching to the latest flow cells and library preparation kits may enhance genotyping accuracy. These observations may point out that, in clinical settings, it may be beneficial to conduct replicates during the initial phase to gather sufficient data for assessing genotyping consistency and intra-sample variability before deciding to switch to genotyping based on a single sample. Furthermore, the R9.4 flow cells and associated sample preparation reagents were discontinued in 2023-2024, leaving V14 chemistry and R10.4 flow cells as the only available option. The cost of full MEFV gene sequencing for 12 samples on a Flongle flow cell is approximately 30 USD, which is comparable to, and often less expensive than, the 26-mutation qPCR panel commonly used in clinical labs in Armenia. With 1 Gb output (2.5 Gb theoretical output according to the manufacturer), 12-sample multiplexing can yield up to 6,000x sequencing depth per sample. The SQK-NBD114.24 kit supports up to 24-sample multiplexing, which could theoretically provide around 3,000x depth per sample - still adequate for variant detection. Additionally, the SQK-NBD114.96 library preparation kit allows for 96-sample multiplexing, making it a viable option for large-scale screening. However, with smaller sample flows, using 24- or 96-sample multiplexing may significantly increase turnaround time. Finally, compared to the Illumina sequencing platform, our protocol offers greater flexibility in the number of samples required for cost-effective multiplexing.

Our method significantly improves upon the protocol proposed by Schmidt et al. (2022), which uses 8 pairs of primers targeting only the exons of the gene, against only 5 primer pairs to target the full gene. The accuracy of variant calling can be further enhanced by switching the basecalling mode to “super accurate”. Although this would require more computational resources and time, it has been shown to outperform default basecalling settings (Ni et al., 2023).

The small sample size is the main limitation of our study. Furthermore, the amount of DNA samples was not sufficient to run NS for all samples in replicates. Nonetheless, our findings, along with previous results (Schmidt et al., 2022), warrant the applicability of full-gene nanopore sequencing of MEFV in clinical settings. Furthermore, even with this small sample, we were able to collect patients with diverse genotype combinations (heterozygotes for a single allele, compound heterozygotes, and homozygotes), which could mitigate potential sampling bias.

In conclusion, we present a cost-effective nanopore sequencing-based protocol for MEFV gene sequencing, making it particularly suitable for settings with limited laboratory infrastructure. It can enable widespread and accurate detection of FMF-related mutations, thus supporting public health initiatives aimed at controlling the prevalence of FMF, particularly in LMICs of endemic regions. This protocol could also lead to the discovery of novel pathogenic variants within the MEFV gene, expanding our knowledge of FMF genetics.