Researchers in Hebei, China, have compiled a detailed, openly available dataset documenting the clinical, pathological, and molecular features of second primary colorectal cancer in cancer survivors, with a particular focus on microsatellite instability. As described in a FAIR²-certified data article published in Frontiers in Oncology, the resource contains structured records from 77 consecutive patients treated at a single hospital-based oncology follow-up program, all diagnosed with a new primary colorectal cancer after surviving a different first malignancy. The dataset’s central contribution is uniform, laboratory-verified documentation of mismatch repair and microsatellite instability status, a molecular phenotype that shapes both prognosis and eligibility for immune checkpoint inhibitor therapy, yet remains poorly characterized in second primary colorectal tumors.
Second primary malignancies are a growing clinical concern as cancer survival improves, and colorectal cancer is among the most common of these subsequent tumors. When a survivor of, for example, breast or lung cancer develops a colorectal tumor years later, clinicians face difficult questions about whether the tumor is biologically distinct from colorectal cancers arising in patients without prior cancer, and whether prior chemotherapy or comorbidity burden influences its location, histology, or molecular profile. Microsatellite instability-high status, typically reflecting defective DNA mismatch repair, is a key molecular phenotype in colorectal cancer, relevant to tumor biology, immunogenicity, and response to immune checkpoint inhibition, but robust data on this phenotype specifically in second primary tumors have been scarce. The new dataset was designed to fill that gap with systematically curated, traceable records.
The cohort was assembled retrospectively from diagnoses made between May 2019 and March 2025. Of 82 patients initially identified with a second primary colorectal cancer, five were excluded: one because of a known hereditary colorectal cancer syndrome, three because of irretrievable missing or ambiguous data, including two with unresolved mismatch repair status, and one because the tumor could not be clearly distinguished as a separate primary. The remaining 77 patients formed the final analytic sample, each confirmed by histopathology and endoscopic review, with complete clinicopathological documentation including definitive mismatch repair and microsatellite instability status. The study received institutional review board approval with a waiver of individual consent given its retrospective, non-interventional design, and it followed the STROBE reporting guidelines for observational studies.
Determining microsatellite instability status was a multi-step laboratory effort. Formalin-fixed, paraffin-embedded tumor sections of 4 micrometers were stained by automated immunohistochemistry for the four key mismatch repair proteins: MLH1, MSH2, MSH6, and PMS2. Internal positive controls, including non-neoplastic crypt epithelium, intratumoral lymphocytes, and stromal cells, had to show unambiguous nuclear staining before a slide could be scored. Complete absence of staining in tumor cells, with intact internal controls, was interpreted as loss of expression; heterogeneous or focal patterns were classified as retained but flagged for independent secondary review. Equivocal or technically suboptimal cases underwent repeat sectioning, restaining, and evaluation by a second gastrointestinal pathologist, with reclassification requiring consensus. When immunohistochemistry could not resolve status, polymerase chain reaction assays targeting a pentaplex panel of mononucleotide markers, BAT-25, BAT-26, NR-21, NR-24, and NR-27, were performed on the original tumor samples, with instability at two or more markers defining MSI-high status.
Across the 77 cases, microsatellite instability-high or mismatch repair deficient status was documented in 16 patients, or 20.8 percent. The anatomical distribution of the second primary tumors was predominantly distal: 27 cases were rectal cancers, 22 were sigmoid colon cancers, 13 were ascending colon cancers, 10 were descending colon cancers, and 5 were transverse colon cancers, yielding 50 colon cancers and 27 rectal cancers overall. Histologically, 65 tumors were conventional adenocarcinomas, 9 were carcinomas in situ, and 3 were mucinous adenocarcinomas. Distant metastasis was present at diagnosis of the second primary in 18 patients, 23.4 percent of the cohort.
The patients themselves represented a diverse survivor population. Nineteen distinct first primary malignancy types appear in the dataset, led by breast cancer with 13 cases, lung cancer with 10, colon cancer with 9, gastric cancer with 7, esophageal cancer with 6, rectal cancer with 5, prostate cancer with 5, bladder cancer with 5, endometrial carcinoma with 4, and renal cell carcinoma with 3. The cohort included 48 men and 29 women. The mean age at first cancer diagnosis was 63.9 years, rising to a mean of 68.7 years at diagnosis of the second primary colorectal cancer. Time between the two diagnoses ranged from zero to 359 months, with a mean of 60 months; 25 patients, 32.5 percent, had synchronous presentation, meaning the interval was zero months, while 52 patients, 67.5 percent, had metachronous tumors diagnosed later.
Comorbidity and treatment data add further context. Hypertension was present in 28 patients, 36.4 percent, diabetes mellitus in 16, or 20.8 percent, and hyperlipidemia in 6, or 7.8 percent, while 35 patients, 45.5 percent, had a documented smoking history. Prior chemotherapy for the first malignancy had been administered to 46 patients, or 59.7 percent. The dataset records these exposures as binary indicators without information on dose, duration, or severity, a constraint the authors flag explicitly for downstream users.
On the analytical side, the study team applied a structured statistical protocol. Continuous variables were summarized as medians with interquartile ranges, and categorical variables as frequencies and percentages. Univariate associations between microsatellite instability-high status and each clinicopathological variable were assessed with two-sided Fisher exact tests for categorical descriptors and Mann–Whitney U tests for continuous data, supplemented by odds ratios, 95 percent confidence intervals, and Cramér’s V effect sizes. Variables reaching the significance threshold, along with selected clinical covariates including hypertension, gastrointestinal first primary cancer, and age at second diagnosis, were entered into a multivariable logistic regression model fitted by maximum likelihood, with multicollinearity assessed by variance inflation factors that did not exceed 3.5. Because only 16 patients had microsatellite instability-high tumors, the model has a low number of events per variable, and the authors caution that its estimates should be regarded as exploratory rather than definitive. Pre-specified sensitivity analyses examined associations between anatomical location and microsatellite instability-high status across subgroups defined by metachronous status, age at second diagnosis, prior chemotherapy, and first cancer type, using Fisher exact tests and recalculated odds ratios.
The dataset’s limitations are candidly documented. Its 77 patients come from a single institution in Hebei Province, China, limiting statistical power and generalizability to other populations and healthcare settings; most subgroups by first primary cancer type include fewer than 15 patients. Complete-case selection, which excluded patients with unresolvable missing data, means the released dataset has no missing values but may carry selection bias, since excluded patients could differ systematically in completeness of workup or patterns of care. Clinically important variables such as TNM stage, tumor grade, detailed treatment regimens, recurrence, and survival are not included, so the resource cannot support prognostic or time-to-event analyses. Molecular data are restricted to mismatch repair and microsatellite instability status; markers such as BRAF V600E, RAS mutations, and MLH1 promoter methylation are absent, and because patients with known hereditary syndromes were excluded while germline testing results are not recorded, sporadic microsatellite instability-high tumors cannot be fully separated from possibly unrecognized hereditary cases. The authors also note that the observed interval between cancers may reflect surveillance intensity, referral patterns, and diagnostic workup after the first cancer, and that inter-rater agreement for data abstraction was not formally quantified.
Despite these constraints, the resource is designed for serious secondary use. It is FAIR²-certified, released under the Open Data Commons Attribution License v1.0, and accompanied by machine-actionable metadata, structured provenance tracking, and a formal data dictionary conforming to standards such as MLCommons Croissant, schema.org, and W3C PROV-O. All 77 records are complete for every one of the dataset’s variables, and each field can be traced to its methodological origin. The authors position the dataset as best suited to descriptive, exploratory, and hypothesis-generating analyses, and to pooling with larger multi-institutional cohorts, rather than stand-alone development of predictive or machine-learning models, which would carry a high risk of overfitting. For researchers studying the epidemiology of multiple primary malignancies, biomarker correlates in sequential cancers, or the integration of molecular phenotyping into survivorship research, the dataset offers a transparent, reproducible starting point built on one of the less frequently captured populations in oncology: patients whose colorectal cancer arrives only after they have already survived another cancer.
Subject of Research: Clinicopathological and molecular characteristics, including MSI-H status, of second primary colorectal cancer in cancer survivors
Article Title: Clinicopathological and molecular characteristics of second primary colorectal cancer in cancer survivors including MSI-H status and anatomical distribution
Article References: Liu, Y., Duan, X., Yang, S., Zhang, Y., & Han, S. (2026). Clinicopathological and molecular characteristics of second primary colorectal cancer in cancer survivors including MSI-H status and anatomical distribution. Frontiers in Oncology, 16, Article 1952368. https://doi.org/10.3389/fonc.2026.1952368
Image Credits: AI Generated
DOI: 10.3389/fonc.2026.1952368
Keywords: colorectal cancer, second primary cancer, microsatellite instability, mismatch repair, MSI-H, cancer survivors, FAIR data, immunohistochemistry, oncology dataset, tumor anatomy, comorbidities, retrospective cohort
News Source: Nathaniel Bowman. (October 23, 2026). New Dataset Profiles Microsatellite Instability in Colorectal Cancers That Strike Cancer Survivors. Scienmag.
