Molecular diagnostic platforms for mitochondrial diseases by conventional and long-read sequencing: Missing pieces

Dear Editor,

The study by Tsai et al. addresses an important clinical need by establishing a molecular diagnostic pipeline for primary mitochondrial diseases (PMDs), comparing conventional RFLP-PCR, Sanger sequencing and third-generation nanopore sequencing [1]. While the authors provide a practical approach in the study, several methodological concerns warrant careful consideration.

Firstly, in Fig. 3 [1], which simulates different heteroplasmy levels of the m.8993T > G mutation that causes Neuropathy, Ataxia and Retinitis Pigmentosa (NARP), the Sanger sequencing electropherogram tracings instead display a T > C transition at the target position. The authors should clarify whether this represents a labelling error or whether the validation experiments were conducted using the T > C variant.

Secondly, sequencing depth or coverage information for the nanopore (long-read) sequencing platform was not described. The sequencing quality indicators such as target region coverage, read depth and read length statistics are important to assess sequencing quality. If long-read sequencing depth is insufficient, there is a substantial risk of false-negative results, particularly for low-heteroplasmy variants. A recent study by Slapnik et al. evaluating nanopore sequencing for mtDNA heteroplasmy detection identified a limit of detection of 12% heteroplasmy and demonstrated that detection capability is highly dependent on sequencing depth [2]. The authors established that Sanger sequencing detected heteroplasmy down to 10–15% but did not assess the analytical detection limit for the nanopore platform [1]. All 22 patients tested with nanopore sequencing returned negative results. Although these results should be subject to clinical correlation, the complete absence of any positive detection raises concerns about the analytical sensitivity of the protocol. A diagnostic test that has not been tested on known positive clinical samples may be considered inadequately validated to detect the conditions for which it is intended. Further validation with larger sample size is needed to determine the sensitivity and specificity of the method tested.

Furthermore, the study might have underutilized the inherent capacity of long-read sequencing technology to sequence very long DNA fragments in a single read. The authors sequenced PCR products of a few kilobases in length, thereby negating the key benefits of long-read technology. Technically it is plausible to sequence native (unamplified) mtDNA or a long-range PCR product spanning the full mitochondrial genome to fully leverage the platform's capabilities [3].

Finally, the distinction between next-generation sequencing (NGS) and third-generation long-read sequencing is not clearly delineated in the title. It is acknowledged that long-read sequencing may be classified within the broad definition of NGS [4]. However, nowadays the nanopore platform is typically defined as a third-generation technology which generates long read sequences while NGS usually refers to massively-parallel sequencing with short reads. The two platforms differ substantially in technical profiles, coverage requirements and validation protocols. Clarifying this distinction would be helpful for readers.

In summary, we agree with the authors’ conclusion that RFLP and Sanger sequencing is an accurate method for PMDs screening, particularly for the 8 hotspots tested. However, the validation of nanopore sequencing without reporting essential sequencing metrics, establishing heteroplasmy detection limits and testing known positive samples represents a significant gap in molecular diagnostic setting.

Declaration of competing interest

None declared.