Real-time quantitative polymerase chain reaction, or RT-qPCR, remains one of the most widely used techniques in breast cancer research. It allows scientists to quantify precisely how active particular genes are in a sample, which is critical for identifying biomarkers, validating molecular targets, and understanding how tumors differ from healthy tissue at the transcript level. But the raw output of an RT-qPCR experiment—the cycle threshold, or Ct value at which fluorescence crosses a detection limit—depends not only on the target gene but also on how much starting material was loaded, how efficiently the RNA was extracted, and how well the reverse transcription step worked. Normalization is the mathematical correction that strips away these technical variables, and it is typically done by measuring a reference gene assumed to be expressed at a constant level in every sample.

The trouble is that this assumption of constancy is often wrong. Genes chosen decades ago as ‘housekeeping’ controls because of their essential roles in basic cellular metabolism can vary substantially between tumor and normal tissue, between molecular subtypes, and even between individual patients. If the reference gene itself fluctuates with disease state, the apparent expression changes in the genes under study may be artifacts of the normalization rather than real biology. The MIQE guidelines, published in 2009, formalized the requirement that researchers validate their reference genes for every specific tissue and experimental context rather than borrowing them uncritically from the literature, and a growing body of work has shown that conventionally used controls frequently fail this test in human cancer studies.

To address this gap for breast tissue, the Moroccan team analyzed twenty breast tissue samples—thirteen tumors and seven normal adjacent tissues—collected at the Military Hospital of Instruction Mohammed V in Rabat, with ethical approval from the Ethics Committee for Biomedical Research in Rabat and written informed consent from all participants. They measured the expression of ten candidate reference genes spanning a range of cellular functions: ACTB, which encodes beta-actin, a core component of the cytoskeleton; GAPDH, a central enzyme of glycolysis; GUSB, a lysosomal enzyme; HNRNPL, an RNA-binding protein involved in splicing; PCBP1, another RNA-binding protein; PPIA, a peptidyl-prolyl isomerase; PUM1, a translational regulator; RER1, an endoplasmic reticulum sorting receptor; TBP, the TATA-box binding protein central to transcription initiation; and 18S ribosomal RNA.

The methodological heart of the study lay in how stability was assessed. Rather than relying on a single algorithm, the researchers used the RefFinder tool, which integrates four independent approaches: geNorm, which ranks candidates based on the pairwise variation between them and identifies the optimal number of genes to combine; NormFinder, a model-based method that estimates both intra- and inter-group variance and is particularly good at accounting for differences between tumor and normal samples; BestKeeper, which uses pairwise correlations of raw Ct values to identify candidates that co-vary least with the rest of the panel; and the comparative delta-Ct method, which compares each candidate directly against all others within each sample. Each algorithm produces its own ranking, and RefFinder condenses these into a geometric mean ranking that reflects the consensus across methods.

The results were strikingly consistent. TBP emerged as the most stable reference gene overall, with a RefFinder geometric mean of 2.21, followed closely by PUM1 at 2.34 and RER1 at 2.78. The three genes occupied closely aligned positions across the individual algorithms, which strengthens confidence in the ranking, since agreement between statistically independent methods is a hallmark of genuine stability. At the other end of the spectrum, ACTB and GAPDH—the two genes most deeply entrenched in the qPCR literature—consistently ranked as the least stable candidates across all four algorithms. In other words, the genes most likely to be chosen by default were the ones most likely to introduce systematic error into a breast tissue expression study.

The finding that GAPDH is unstable in breast tumors is biologically plausible rather than merely statistical. Tumor cells undergo a metabolic shift toward aerobic glycolysis, the Warburg effect, and GAPDH sits at the center of that reprogramming, so its expression can rise with tumor aggressiveness and hypoxia. ACTB, meanwhile, is tied to cytoskeletal remodeling, a process intimately involved in invasion and metastasis. TBP, by contrast, is a basal transcription factor whose expression tends to track overall cellular transcriptional activity more evenly, and PUM1 and RER1 have shown stable expression in other cancer contexts, including rectal tumors and cancer cell line panels, lending external support to the new results.

The study also fits into a broader pattern in the normalization literature. Previous work in breast cancer cell lines, including the widely used MCF-7 line, has identified different optimal reference sets than those suited to fresh tissue, and studies in endometrial, ovarian, colorectal, and renal cancers have each concluded that no single reference gene is universally appropriate. Some groups have advocated normalizing to the geometric average of multiple reference genes rather than a single control, an approach pioneered by Vandesompele and colleagues in 2002, and the close agreement among TBP, PUM1, and RER1 in this study suggests that a combination of two or three of them could provide a robust normalization strategy for breast tissue work.

The authors are appropriately measured about the limitations of their work. The sample size of twenty tissues, while sufficient for a candidate-gene validation study of this design, is modest, and the tumor cohort of thirteen samples cannot capture the full heterogeneity of breast cancer, which spans estrogen receptor-positive, HER2-positive, and triple-negative subtypes with distinct molecular landscapes. The team explicitly calls for further validation on a larger scale before the findings are adopted as standard practice. They also note that the datasets generated in the study are available from the corresponding author on reasonable request, and that all summarized data supporting the conclusions are included in the published article, in keeping with transparency norms in the field.

Even with those caveats, the practical implications are immediate for laboratories working on breast cancer gene expression. Any study planning to use RT-qPCR on breast tumors or matched normal adjacent tissue now has a data-driven, multi-algorithm-validated shortlist of candidates—TBP, PUM1, and RER1—and a clear warning against defaulting to ACTB or GAPDH. As RT-qPCR continues to complement immunohistochemistry in molecular subtyping and to support the validation of therapeutic targets such as PIK3CA and BRCA1/2 in the era of precision oncology, the accuracy of the underlying normalization quietly underpins every downstream conclusion. A study like this one, though modest in scale, addresses one of the most consequential and least glamorous sources of error in cancer molecular biology, and it does so with the kind of methodological rigor that the MIQE guidelines were written to encourage.

Subject of Research: Validation of reference genes for RT-qPCR normalization in breast tumors and normal adjacent tissues

Article Title: Promising reference genes for RT-qPCR normalization in breast tumors and normal adjacent tissues

Article References: Oubaqui, F.-E., Oukabli, M., Kouach, J., Bakri, Y., Ameziane El Hassani, R., & Qmichou, Z. (2026). Promising reference genes for RT-qPCR normalization in breast tumors and normal adjacent tissues. Molecular Biology Reports, 53(1), Article 1646. https://doi.org/10.1007/s11033-026-12851-2

Image Credits: AI Generated

DOI: 10.1007/s11033-026-12851-2

Keywords: reference genes, RT-qPCR, breast cancer, gene expression normalization, TBP, PUM1, RER1, GAPDH, ACTB, housekeeping genes, molecular biomarkers, MIQE guidelines

The research team, led by Yuanxuan Xia of the Department of Neurosurgery at Johns Hopkins University School of Medicine, screened 1,401 patients who underwent spinal MRI between 2020 and 2025 at a tertiary academic center. From that pool, only 69 patients met the strict inclusion criteria: they had intramedullary lesions, which arise within the spinal cord itself, or leptomeningeal lesions, which seed the delicate membranes surrounding the cord, and they had a primary cancer that originated outside the central nervous system. Patients whose tumors began in the brain or spinal cord, and those with non-cancer diagnoses, were excluded to ensure the cohort represented true metastatic spread from systemic disease.

The composition of the cohort reflects the broader epidemiology of CNS metastasis. Lung cancer accounted for 33.3 percent of primary tumors, followed closely by breast cancer at 31.9 percent, with the remainder drawn from other solid malignancies. Only 11.6 percent of patients underwent surgery for their intradural disease, underscoring how often these lesions are managed with radiation, systemic therapy, or palliative care rather than operative intervention. The anatomical distribution of disease at first detection was striking: 55.1 percent of patients had cervical involvement, 65.2 percent thoracic, and 76.8 percent lumbar, suggesting that tumor cells descending through the cerebrospinal fluid tend to accumulate in the lower reaches of the thecal sac, where gravity and CSF flow patterns favor deposition.

Perhaps the most consequential finding of the study is its reconstruction of the metastatic timeline. Among the 69 patients, the median interval from primary cancer diagnosis to any metastasis was just 0.5 months, with an interquartile range spanning 0 to 6.7 months, indicating that systemic dissemination often begins almost immediately, sometimes even before the primary tumor is detected. Spread to the brain followed at a median of 14.9 months, while seeding of the osseous spine, the bony vertebrae surrounding the cord, occurred at a median of 19.4 months. Intradural spinal involvement, by contrast, was a distinctly late event, arriving at a median of 29.5 months after the primary diagnosis, with an interquartile range of 14.0 to 52.6 months.

This chronology carries important biological implications. It suggests that tumor cells capable of surviving in the cerebrospinal fluid and invading the intradural compartment represent a late-arising, highly selected subclone within the evolving cancer population. The intradural space is immunologically and pharmacologically distinct from the rest of the body: it is sheltered by the blood-spinal cord and blood-brain barriers, bathed in CSF, and relatively inaccessible to many conventional chemotherapies. Cells that successfully colonize it must acquire a suite of adaptations, from altered adhesion molecules to enhanced survival signaling in a nutrient-poor environment, and the long delay before their emergence is consistent with the gradual accumulation of these traits over successive rounds of dissemination and selection.

The prognosis attached to this late stage is sobering. Within a median follow-up of 42.0 months after primary cancer diagnosis, the mortality rate in the cohort reached 87.0 percent. That figure, drawn from an interquartile range of 20.6 to 72.9 months of follow-up, reflects both the aggressiveness of the underlying cancers and the limited therapeutic arsenal available once disease has penetrated the intradural compartment. Leptomeningeal metastasis in particular has long been associated with poor outcomes across solid tumors, and the present data reinforce the sense that intradural involvement marks an advanced chapter in the metastatic story rather than an isolated complication.

Yet the study also uncovered a genuine therapeutic opening in its molecular analysis. Targetable mutations, alterations in genes for which approved or investigational targeted drugs exist, were identified in 46.4 percent of primary lesions. Patients whose tumors carried such biomarkers trended toward improved overall survival, with a median of 51.7 months compared with 30.1 months for those without actionable alterations, a difference that approached but did not reach conventional statistical significance at p equal to 0.072. The critical distinction emerged when the researchers separated patients who actually received targeted therapies from those who did not. Among patients with actionable biomarkers who qualified for and received matched targeted treatment, median overall survival rose to 56.2 months, significantly better than the 31.2 months observed in those who did not receive such therapy, with p equal to 0.029.

That survival gap, though drawn from a small and retrospective cohort, aligns with a growing body of evidence that molecular profiling of CNS metastases can change the trajectory of disease. Prior work has shown that genomic data obtained from spinal metastatic tumor samples are clinically reliable, and recent multicenter studies of leptomeningeal metastasis in non-small cell lung cancer have demonstrated that cerebrospinal fluid circulating tumor DNA profiling can stratify risk and guide matched treatment. The Johns Hopkins findings extend this logic to the intradural spinal compartment specifically, arguing that nearly half of patients who develop disease there carry druggable alterations detectable in their primary tumors.

The study’s authors conclude that regular molecular characterization and sequencing of the intradural space, for example through CSF liquid biopsy, could have a clinically meaningful impact on these patients. Liquid biopsy of cerebrospinal fluid offers a minimally invasive window into the genetic landscape of tumors sheltering behind the blood-CNS barriers, capturing shed DNA that peripheral blood draws often miss. As sequencing technologies become smaller, cheaper, and faster, the prospect of routinely sampling CSF in patients with known metastatic disease becomes increasingly practical, potentially allowing clinicians to detect intradural seeding earlier in the cascade and to select targeted agents before neurological function is compromised.

The broader significance of the work lies in its framing of intradural spinal metastasis not as a random complication but as a definable, time-stamped stage within the metastatic cascade. By quantifying when in the disease course these lesions appear, where they settle along the spinal axis, and what molecular vulnerabilities they inherit from their primary tumors, the study provides clinicians with a roadmap for surveillance and a rationale for aggressive molecular workup. For a patient population in which median survival is measured in years rather than decades, and in which nearly nine in ten patients do not survive beyond the median follow-up window, any intervention that adds months of meaningful life, as targeted therapy appears to do, represents progress worth pursuing. The challenge ahead is to translate this retrospective portrait into prospective strategies: earlier CSF-based screening for high-risk patients, systematic sequencing of both primary tumors and intradural deposits, and clinical trials of targeted and CNS-penetrant agents designed specifically for this late but biologically distinct frontier of cancer spread.

Subject of Research: Clinical course, dissemination timing, and molecular features of intradural spinal metastasis in patients with systemic cancer

Article Title: Characterizing intradural spinal metastasis in the metastatic cascade: clinical course, dissemination timing, and molecular features in 69 patients

Article References: Characterizing intradural spinal metastasis in the metastatic cascade: clinical course, dissemination timing, and molecular features in 69 patients. (n.d.). https://doi.org/10.1007/s11060-026-05794-y

Image Credits: AI Generated

DOI: 10.1007/s11060-026-05794-y

Keywords: intradural spinal metastasis, leptomeningeal metastasis, metastatic cascade, spinal cord, cerebrospinal fluid, liquid biopsy, targeted therapy, lung cancer, breast cancer, molecular biomarkers, overall survival, neuro-oncology