One Gene, Two Proteins: Why CFH Cancer Biomarker Studies May Be Measuring the Wrong Molecule

A single gene has been quietly generating a great deal of excitement in cancer research, and a new commentary published in BMC Cancer argues that much of that excitement rests on a measurement problem that most laboratories have never thought to check. The gene is CFH, which encodes complement factor H, a master regulator of the complement system, the ancient branch of innate immunity that normally tags microbes and dying cells for destruction. Over the past several years, large pan-cancer analyses have repeatedly found that CFH transcription is dysregulated across many tumour types and that its expression correlates with clinical stage, patient survival, immune cell infiltration, tumour mutational burden, and microsatellite instability. On the face of it, CFH looks like exactly the kind of immune-related prognostic biomarker that oncology has been searching for. But according to Simon J. Clark of Eberhard Karls University of Tübingen and the University of Manchester, writing in a Matters Arising piece, the standard way of measuring CFH activity in these studies conflates two fundamentally different proteins, and that conflation could matter enormously when the findings are translated into diagnostic tests.

The source of the ambiguity lies in the peculiar architecture of the CFH gene itself. From a single genetic locus, cells produce two distinct protein products through alternative splicing. The predominant product is full-length complement factor H, usually abbreviated FH, a large glycoprotein built from twenty complement control protein domains that circulates in plasma and patrols host surfaces, distinguishing self from threat and restraining the alternative pathway of complement activation. The second product is factor H–like 1, or FHL-1, a much smaller splice isoform that consists only of the first seven of those twenty domains, followed by a short, unique C-terminal tail generated by the splicing event. FHL-1 retains the core complement regulatory activities of its larger sibling, including its cofactor function for factor I and its ability to regulate C3b deposition, but it lacks the C-terminal host-recognition domains that allow full-length FH to anchor itself to cell surfaces. In structural and functional terms, these are not interchangeable molecules; they are two tools with overlapping but clearly distinct jobs.

What makes FHL-1 far more than a molecular footnote is where it ends up in the body. It is abundantly expressed in several tissues, and tumour cells themselves, including lung and pancreatic cancer cells, have been shown to produce it, contributing to local complement regulation within the tumour microenvironment. Crucially, FHL-1 exhibits enhanced tissue retention through interactions with the extracellular matrix, the dense protein scaffold that surrounds and structures solid tumours. Because it is smaller than full-length FH and is not glycosylated, FHL-1 can diffuse into and penetrate tightly packed extracellular matrices in a way that the bulky, sugar-decorated full-length protein cannot. Clark points to previous work in the eye, where FHL-1 was identified as the predominant complement regulator in Bruch’s membrane, a densely cross-linked extracellular layer, as evidence of this penetration phenomenon in human tissues. In glioblastoma, meanwhile, researchers have observed increases specifically in FHL-1 expression and secretion, with the hypothesis that the small isoform’s diffusibility allows it to infiltrate the tumour’s matrix-rich architecture.

Against this biological backdrop, the technical critique of the recent pan-cancer analysis becomes easy to understand. The original study by Zheng and colleagues, published in BMC Cancer in 2025, quantified CFH transcription using quantitative reverse-transcription PCR with primers targeting the N-terminal region of the gene. Those primers bind to exons that are shared by both the FH and FHL-1 transcripts. The signal they produce therefore reflects the combined transcription of both isoforms, not full-length factor H alone. The same problem afflicts the functional experiments: the short hairpin RNA constructs used to knock down CFH in pancreatic cancer cells target shared N-terminal sequences, so they would be expected to suppress both FH and FHL-1 simultaneously. The proliferation and migration defects observed after knockdown, in other words, cannot be attributed to either isoform individually. The transcriptional, prognostic, and functional associations reported in the study represent an aggregate FH-plus-FHL-1 signal, without any isoform-specific resolution.

Why should this matter to anyone outside the complement field? The answer lies in how biomarkers are built and validated. Clinical biomarker development depends on clear molecular definition and reproducibility across platforms, assays, and patient cohorts. If FH and FHL-1 are differentially expressed across tumour types, disease stages, or microenvironmental contexts, then an assay that cannot tell them apart will yield signals that are biologically heterogeneous, mixing together two molecules that may behave very differently in different patients. A correlation that holds for the blended signal might be driven almost entirely by one isoform in one cancer and by the other isoform in a different cancer, or it might shift as a tumour progresses and remodels its extracellular matrix. Without isoform resolution, the underlying biology remains invisible behind a single composite number.

The stakes rise further when the two isoforms are considered as potential drivers of tumour biology rather than passive correlates. Because FHL-1 preferentially localizes within the extracellular matrix, it may disproportionately influence local complement activity and immune cell behaviour inside solid tumours, precisely where the battle between tumour cells and immune effectors is fought. Complement is increasingly recognized as a modulator of the tumour microenvironment, shaping inflammation, immune suppression, and the response to immunotherapy. Consequently, correlations reported between aggregate CFH expression and immune infiltration, prognosis, or therapeutic responsiveness may, in some cancers, be driven largely by FHL-1 rather than by full-length FH. If that is the case, the mechanistic story inferred from the data, and the therapeutic strategies built upon it, could be pointed in the wrong direction.

The implications extend directly into the clinic. When CFH transcription is proposed as a diagnostic or prognostic biomarker, or as a marker to guide immunotherapeutic stratification, isoform-specific differences could affect the sensitivity, specificity, and predictive value of FH-based assays. They could also complicate the interpretation of measurements taken from different compartments, since circulating blood levels and tissue-derived levels may reflect different balances of the two proteins. A blood test capturing mostly circulating full-length FH and a tissue assay dominated by matrix-retained FHL-1 could give contradictory answers about the same patient, and clinicians would have no way of knowing which molecule they were actually tracking. In an era when biomarker-driven trials can succeed or fail on the precision of patient selection, such ambiguity is not a trivial technicality.

Importantly, Clark is careful to frame the commentary as a constructive refinement rather than a demolition. He emphasizes that the observation does not detract from the significance or rigour of the original pan-cancer analysis, which he describes as comprehensive and high-quality, integrating transcriptomic data from large public datasets with immune infiltration analyses and in vitro functional validation. That study, he writes, provides a strong foundation for future biomarker-oriented investigations into complement regulation in cancer. The critique is about the next step: studies aimed at clinical translation would benefit from incorporating isoform-resolved approaches before CFH-based signatures are locked into diagnostic pipelines. The tools to do so already exist, including splice-junction–specific quantitative PCR that can distinguish the unique FHL-1 tail from shared exons, long-read RNA sequencing capable of resolving full-length transcript structures, and proteomic methods able to separate the two proteins by mass and by their distinctive C-terminal peptides.

For the broader cancer research community, the episode is a reminder of a principle that applies well beyond complement genetics. Pan-cancer analyses, for all their statistical power, are only as specific as the molecular probes on which they rest, and genes that produce multiple protein products through alternative splicing are a persistent blind spot. The CFH locus is far from unique in this respect, and as transcriptomic biomarkers move from discovery papers toward clinical deployment, the demand for isoform-aware assay design will only grow. Clark’s suggestion is straightforward: refine the measurement, and the clinical utility of FH-based biomarkers may sharpen accordingly, clarifying at the same time how complement regulatory pathways can genuinely be leveraged for cancer diagnosis, prognosis, and therapeutic stratification. The difference between a promising correlation and a deployable diagnostic, it turns out, may hinge on seven protein domains and a short, unique tail.

Subject of Research: Isoform-specific interpretation of CFH gene expression as a pan-cancer biomarker

Article Title: Diagnostic interpretation of CFH expression in pan-cancer analyses: the importance of factor H–like 1 isoform resolution

Article References: Clark, S. J. (2026). Diagnostic interpretation of CFH expression in pan-cancer analyses: the importance of factor H–like 1 isoform resolution. BMC Cancer, 26(1), Article 1214. https://doi.org/10.1186/s12885-026-17134-4

Image Credits: AI Generated

DOI: 10.1186/s12885-026-17134-4

Keywords: CFH, complement factor H, FHL-1, biomarkers, pan-cancer analysis, alternative splicing, tumour microenvironment, complement system, extracellular matrix, prognosis, immunotherapy, qRT-PCR