The study, published in Clinical Cancer Bulletin, drew on one of the largest single-center myeloma cohorts in China, spanning patients diagnosed between January 2010 and April 2021. Of the 812 patients enrolled, 703 did not undergo autologous stem cell transplantation and were randomly split into a training set of 527 and a validation set of 176, while 109 transplant recipients formed a separate group. All patients received guideline-consistent first-line therapy built around proteasome inhibitors, immunomodulators, or both, ensuring that differences in outcome could be attributed to disease biology rather than wildly different treatments.

What sets the study apart is its breadth. Rather than focusing narrowly on the cytogenetic abnormalities that dominate contemporary myeloma research, the investigators collected everything from hemoglobin and platelet counts to complement proteins, interleukin levels, anion gap, and the patient’s overall functional status as measured by the Eastern Cooperative Oncology Group (ECOG) scale. Missing data, which affected fewer than 15 percent of values for any parameter, were handled with multiple imputation by chained equations, a statistical technique that generates several plausible completed datasets and pools the results to avoid bias.

In the first pass, univariate Cox regression analysis of the training set identified 29 parameters with measurable prognostic value when treated as continuous variables. When the researchers converted those variables into clinically practical categories, using thresholds such as the International Myeloma Working Group criteria for anemia, renal insufficiency, and hypercalcemia, 25 parameters retained their predictive power. Notably, some familiar markers lost their luster in this transformation: age, M protein quantification, and alkaline phosphatase no longer distinguished survivors from non-survivors once dichotomized, while urea nitrogen emerged as newly informative.

The decisive filtering came with multivariate Cox regression, which tests whether each parameter adds predictive value independently of all the others. Ten factors survived: ECOG performance score, the presence of extramedullary lesions, platelet count, reticulocyte count, anion gap, hypercalcemia, complement C3, beta-2 microglobulin, high-risk cytogenetics, and interleukin-2 receptor levels. Some of these, such as beta-2 microglobulin and cytogenetic abnormalities, are already embedded in mainstream staging systems. Others, including the anion gap, a routine electrolyte-derived measure, and the reticulocyte count, a marker of bone marrow’s regenerative capacity, have been almost entirely overlooked in myeloma prognostication.

Several of these findings challenge conventional wisdom. Albumin, a cornerstone of the International Staging System for two decades, showed prognostic value in univariate analysis but failed to hold up in the multivariate model. The proportion of plasma cells in the bone marrow, often assumed to reflect tumor burden, likewise did not independently predict outcome, which the authors attribute to the disease’s profound internal heterogeneity: how the tumor behaves matters more than how large it is. Meanwhile, 1q21 gain or amplification, an abnormality given weight in the newest R2-ISS staging revision, did not reach statistical significance in this cohort.

The researchers also benchmarked the four widely used staging systems, Durie-Salmon, ISS, R-ISS, and R2-ISS, against real-world outcomes using time-dependent receiver operating characteristic curves. The results were sobering. The best performer, R2-ISS, achieved areas under the curve of roughly 0.68 to 0.72 for one-, three-, and five-year overall survival, while the Durie-Salmon system hovered near 0.52 to 0.54, barely better than a coin flip. In a disease where treatment intensity decisions hinge on risk category, that margin of error carries real clinical consequences.

From the ten surviving parameters, the team constructed the Zhongshan Risk Score, a weighted sum in which each factor contributes according to its Cox regression coefficient, with beta-2 microglobulin, platelet count, and hypercalcemia carrying the heaviest weights. Using a cut-off of 2.02, patients in the training set split cleanly into high- and low-risk groups with dramatically different overall and progression-free survival, both differences highly significant. The score held up in the validation set and, importantly, also stratified the 109 transplant patients, suggesting it applies across treatment modalities. Its accuracy was striking: area under the curve values of 0.81, 0.81, and 0.83 for one-, three-, and five-year survival, comfortably exceeding every existing staging system. A companion nomogram translates the score into estimated one-, three-, and five-year survival probabilities for individual patients.

The biology behind the novel markers remains speculative but tantalizing. Complement C3, a central protein of the innate immune system, can activate macrophages and drive complement-dependent cytotoxicity, potentially mirroring the body’s anti-tumor response. Interleukin-2 receptor, by contrast, acts as a negative immunomodulator, and elevated levels have previously been linked to treatment resistance in myeloma. The anion gap’s role is murkier, possibly reflecting metabolic derangement or renal impairment, while reticulocyte and platelet counts appear to capture the residual health of the bone marrow’s normal hematopoietic machinery, a dimension that hemoglobin alone fails to convey.

The authors are candid about limitations. The study is retrospective and single-center, raising questions about generalizability across ethnic groups and health systems, and emerging biomarkers such as circulating tumor DNA were not available for analysis. The transplant subgroup was small, warranting caution. Still, the implications are considerable: a cheap, widely accessible panel of routine blood tests and clinical assessments could sharpen risk stratification beyond what expensive cytogenetic panels achieve alone, potentially guiding which patients warrant early escalation to monoclonal antibodies or CAR T-cell therapy, and which can be spared overtreatment. External validation in multicenter prospective cohorts will determine whether the Zhongshan Risk Score earns a place beside, or above, the staging systems it just outperformed.

Subject of Research: Prognostic clinical parameters and risk stratification in multiple myeloma

Article Title: Comprehensive evaluation of clinical prognostic parameters in a real-world cohort of 812 patients with multiple myeloma

Article References: Wang, Y., Lan, T., Zhou, C., Xu, T., & Liu, P. (2025). Comprehensive evaluation of clinical prognostic parameters in a real-world cohort of 812 patients with multiple myeloma. Clinical Cancer Bulletin, 4(1), Article 4. https://doi.org/10.1007/s44272-025-00031-5

Image Credits: AI Generated

DOI: 10.1007/s44272-025-00031-5

Keywords: multiple myeloma, prognosis, risk score, biomarkers, staging systems, beta-2 microglobulin, cytogenetics, ECOG score, complement C3, interleukin-2 receptor, real-world cohort, hematology