Rafael Parra-Medina

Introduction: Cancer incidence is rising rapidly in low- and middle-income countries (LMICs), where diagnostic pathology capacity is constrained. AI applied to routine histopathology could strengthen diagnostic capacity and offer leapfrogging potential by embedding AI into diagnostic systems as they are built up. However, without empirically identifying implementation determinants in LMIC cancer-diagnostic settings, AI solutions risk being poorly aligned with local constraints, limiting impact or exacerbating existing disparities. Methods: We conducted an exploratory mixed-methods pilot study across three phases: 1) a structured survey characterising the diagnostic and technical landscape and AI priorities; 2) focus group discussions (FGDs) to identify implementation determinants and 3) a second survey quantifying FGD-derived themes. Twelve physicians (nine pathologists, three gynaecologists) were purposively recruited (10 LMIC, 2 high-income countries for reference) and findings integrated through narrative triangulation. Results: Diagnostic and AI infrastructure varied substantially and did not track consistently with country income classification. H&E staining was universally available, immunohistochemistry capacity was below WHO-recommended capacity for cancer diagnosis at 7 of 10 sites and molecular testing absent or minimal at 6 of 10 sites. Whole-slide-imaging scanners were available at 8 of 10 sites, graphics processing units, required for efficient AI inference, were available at only 4 of 10 sites. Participants prioritised AI for cancer screening, diagnosis, and biomarker prediction from H&E slides. Six implementation determinants were identified: AI literacy, governance frameworks, funding, evidence generation, clinical deployment, and collaboration. Financial constraints emerged as the dominant barrier across all determinants, and collaboration as a key enabler. Conclusion: AI implementation readiness varied substantially within income groups, suggesting that country income classification is an insufficient proxy for institutional readiness, underscoring the need for institution-level assessment. Realising the leapfrogging potential of AI in this context requires coordinated progress across all determinants, with sustainable financing, and collaboration as enablers of the other determinants. Key messages: What is already known on this topic?: Cancer incidence is rising rapidly in LMICs, where diagnostic capacity and molecular testing capabilities are severely constrained. AI could alleviate gaps, yet evidence on implementation determinants for integrating AI into cancer diagnostics in these settings remains limited. What this study adds?: In this exploratory sample, AI implementation readiness varied substantially across LMIC settings, including within the same country income group. Readiness is shaped by six implementation determinants: AI literacy, governance frameworks, funding, evidence generation, clinical deployment, and collaboration, with financial constraints as the dominant cross-cutting barrier. How this study might affect research, practice or policy?: These findings support institution-level feasibility assessment rather than reliance on country income classification as a proxy and highlight the need for sustainable financing and collaboration as foundational prerequisites for AI implementation in cancer diagnostics in LMICs.

Background/Objectives: The 2021 World Health Organization Classification of Tumors of the Central Nervous System (WHO CNS5) integrates histological and molecular criteria for adult diffuse gliomas, but comprehensive molecular testing may be difficult to implement in resource-limited settings. This study assessed the feasibility of a sequential, marker-based approach for diagnostic reassessment and described its association with clinical characteristics and overall survival (OS) in adults from Colombia. Methods: We retrospectively analyzed 68 adults with diffuse gliomas treated at two referral hospitals in Bogotá, Colombia, between 2015 and 2023. Molecular assessment used IHC, FISH, PCR, and Sanger sequencing according to the available tissue and diagnostic workflow. The study did not include comprehensive NGS or systematic assessment of combined whole-chromosome 7 gain/chromosome 10 loss (+7/−10). Results: The diagnosis changed after molecular reassessment in 51.5% of cases. The proportion classified as glioblastoma increased from 48.5% under the historical histologic classification to 66.2% under the WHO CNS5-based algorithm, whereas FISH-defined oligodendrogliomas decreased from 17.6% to 11.8%. OS differed among historical histologic groups (p = 0.03), but not among the WHO CNS5-based groups (p = 0.13). Female sex was associated with longer OS in unadjusted analysis (p = 0.02), while age was the only independent predictor in the multivariable analysis. Conclusions: Sequential molecular reassessment materially changed diagnostic categorization in this cohort and was feasible using locally available marker-based methods. However, the results are exploratory, limited by small molecular subgroups, incomplete molecular coverage, and the retrospective treatment context; they should not be interpreted as demonstrating improved treatment selection or survival after reclassification.

Background: Lung adenocarcinoma is the most common subtype of non-small cell lung cancer and a leading cause of cancer-related mortality worldwide. Its marked clinical heterogeneity results in variable outcomes and poor prognosis for many patients. This study aimed to develop and compare statistical, machine learning, and deep learning models for predicting overall survival (OS) in patients with primary lung adenocarcinoma. Methods: This retrospective cohort study included patients with primary lung adenocarcinoma. Missing data were handled using fold-specific multiple imputation by chained equations within a nested cross-validation framework. Predictor selection and hyperparameter tuning were performed within 5 outer and 3 inner folds using univariable Cox analysis, Elastic Net, XGBoost-based importance, and clinical criteria. Overall survival was modeled using penalized Cox regression, Random Survival Forest, DeepSurv, and DeepHit. Model performance was evaluated from out-of-fold predictions using concordance indices, time-dependent AUC, calibration and the integrated Brier score. Sensitivity analyses included propensity score-matched survival comparisons by biomarker status among patients with stage IV disease. Results: The cohort included 254 patients, with a median overall survival of 12.8 months. Random Survival Forest showed the best performance (Harrell’s C-index, 0.760; mean time-dependent AUC, 0.851; integrated Brier score, 0.160) and clearly separated high- and low-risk groups (p < 0.001). Clinical stage, ECOG status, systemic treatment, sex, and age were among the most influential predictors. Calibration was acceptable, with slight survival overestimation. Exploratory propensity score-matched analyses in stage IV subgroups suggested the highest apparent performance for RSF, although the small matched samples limited definitive conclusions. Conclusions: In this cohort, machine learning models showed potential for overall survival prediction in lung adenocarcinoma. RSF achieved the highest observed discrimination, while the penalized Cox model showed competitive performance with lower complexity. These findings suggest that RSF may complement conventional survival models; however, the retrospective single-center design, moderate sample size, missing data, and lack of external validation limit their generalizability.

Background/Objectives: The reclassification of gliomas in the 5th edition of the World Health Organization’s Classification of Tumors of the Central Nervous System (WHOCNS5) integrates histological and molecular criteria to improve diagnostic accuracy and enable an individualized treatment approach. This study aims to analyze the feasibility of WHOCNS5 reclassification by examining its impact on diagnostic distribution, clinical features, and overall survival (OS) in adult patients from Colombia. Methods: We retrospectively analyzed 68 adults with diffuse gliomas (2015–2023) from two referral hospitals in Bogotá, Colombia. Molecular testing for IDH, 1p/19q, ATRX, TP53, EGFR, TERT, and CDKN2A/B enabled tumor reclassification according to WHOCNS5. Clinical, radiographic, and survival data were statistically analyzed. Results: 58.8% of patients were male, with a mean age of 56.6 ± 15.5 years. A total of 51.5% of the cases varied in the diagnosis after molecular reclassification. WHOCNS5 identified 66.2% glioblastomas vs 48.5% according to WHOCNS3, and 11.8% oligodendrogliomas vs 17.6 according to WHOCNS3. Using the WHOCNS3 criteria, OS differed significantly among tumor subtypes (p = 0.03), while reclassification under WHOCNS5 reduced these differences (p = 0.13), although IDH-mutant astrocytomas and oligodendrogliomas showed a survival advantage. In all subtypes, a younger age and gender predicted better OS (p < 0.001 and p = 0.02, respectively). Conclusions: Implementing the WHOCNS5 reclassification substantially altered glioma categorization and refined prognostic interpretation in over half of the cases. Trends toward improved outcomes in IDH-mutant tumors persist after reclassification.

Background: Radioiodine (RAI)-refractory differentiated thyroid cancer (DTC) has a poor prognosis. Constitutive MAPK activation, frequently driven by BRAF V600E or RAS alterations, suppresses thyroid-specific genes and sodium-iodide symporter expression. Short-course BRAF and/or MEK inhibition before RAI aims to restore iodine avidity. Methods: We conducted a systematic review and meta-analysis in accordance with PRISMA 2020. PubMed, Embase, CENTRAL, ClinicalTrials.gov, and Google Scholar were searched from inception through March 31, 2025, for prospective studies of short-course BRAF and/or MEK inhibition undertaken with the protocolized intention of restoring RAI uptake before therapeutic RAI (PROSPERO: CRD1371813). Random-effects models with REML estimation and logit transformation were used for proportions. Risk of bias was assessed with ROBINS-I and certainty with GRADE. Results: Five non-randomized single-arm studies (N=72) met the revised eligibility criteria; four studies (n=48) contributed to the primary meta-analysis. The pooled proportion with restored detectable RAI uptake was 62.3% (95% CI 47.8%-74.8%; I²=0%). When the therapeutic dosimetric threshold reported by Ho et al. was used, the pooled estimate was 57.8% (95% CI 44.0%-71.5%). Excluding Ho et al., it yielded 63.5% (95% CI 45.7%-81.4%). No included study published a hazard ratio for progression-free survival; progression-free and overall survival were therefore synthesized narratively. Adverse-event rates were not pooled because attribution, grading, and exposure duration differed across studies. No treatment-related deaths were reported. Certainty of the evidence was very low for all outcomes. Conclusions: Short-course MAPK inhibitor-based redifferentiation can restore RAI uptake in selected patients with RAI-refractory DTC, but uptake is a surrogate outcome and does not establish survival benefit. The evidence is insufficient to determine whether efficacy differs by mutation subtype. Controlled, biomarker-stratified trials using standardized imaging and clinically relevant outcomes are needed.

Field cancerization describes molecular alterations occurring in histologically normal tissues surrounding tumors that may contribute to cancer initiation and progression. In prostate cancer (PCa), the molecular characteristics of peritumoral tissue (PTT) remain incompletely understood. Because microRNAs (miRNAs) play key roles in gene regulation, tumor progression, and microenvironmental remodeling, we investigated miRNA expression patterns and regulatory networks across benign tissue (BT), PTT, and tumor tissue (TT). Small RNA sequencing was performed on matched formalin-fixed paraffin-embedded samples from 40 patients with PCa. Differential expression analysis was conducted using DESeq2, adjusting for age and Gleason grade, while functional enrichment analysis and weighted gene co-expression network analysis (WGCNA) were used to identify dysregulated pathways and conserved miRNA modules. PTT exhibited a molecular profile intermediate between BT and TT, consistent with a field cancerization effect. Compared with BT, 102 miRNAs were differentially expressed in TT and 57 in PTT, with 39 miRNAs (68% of the PTT-associated miRNAs) overlapping the tumor signature. Shared dysregulated pathways included PI3K–Akt, p53, and HIF-1 signaling; whereas, PTT showed additional enrichment in pathways related to epigenetic regulation (Polycomb Repressive Complex) and cellular stress responses (mitophagy, protein processing in ER) exclusively through up-regulated miRNAs; no pathways were uniquely enriched from down-regulated miRNAs in PTT. WGCNA identified conserved miRNA modules enriched for members of the let-7, miR-200, miR-103/107, and miR-106a~363 families, which have established roles in epithelial–mesenchymal transition, tumor progression, and microenvironmental remodeling. Collectively, these findings demonstrate that histologically benign peritumoral tissues harbor tumor-associated miRNA programs and regulatory networks that closely resemble those observed in prostate tumors, providing molecular evidence of field cancerization in PCa and identifying potential miRNA-mediated mechanisms relevant to disease progression and biomarker development.

Background: Colorectal cancer (CRC) is the third most common cancer globally, with rising cases in Latin America. MSI-H and MMRd play key roles in CRC, but data on their prevalence in Hispanic/Latino populations are limited. This study evaluates these biomarkers in the region. Methods: A systematic review and meta-analysis were conducted following PRISMA guidelinesin Medline, Virtual Health Library, Scopus, and Web of Science. Random-effects models were used to estimate pooled prevalence due to expected heterogeneity between studies. Studies (cohort and cross-sectional) that evaluated MMRd and/or MSI-H through IHC and PCR techniques in Hispanic/Latino individuals with colorectal cancer (whether sporadic, associated with Lynch Syndrome, or other forms), residing in Latin American countries or elsewhere, were included. Results: A total of 52 studies including 10,596 patients were included. The pooled prevalence of mismatch repair deficiency (MMRd) and microsatellite instability-high (MSI-H) in Hispanic/Latino populations was 15% (95% CI: 10%–20%; I ² = 89.6%) and 18% (95% CI: 13%–24%; I ² = 84.0%), respectively. Costa Rica and Mexico had the highest MMRd prevalence (30% and 24%), while Uruguay showed the highest MSI-H prevalence (45%). MSI-H was significantly associated with female sex (OR: 1.83) and right-sided tumors (OR: 8.16). MMRd was associated with right-sided tumors compared with the rectum (OR: 1.73) and the left colon (OR: 5.65). Conclusions: This meta-analysis underscores the unique prevalence of MMRd and MSI-H in Hispanics, highlighting regional variations and the need for broader representation.

Introduction: metastases to the thyroid gland are a rare finding. They occur more frequently in women and may be synchronous or metachronous, by dissemination from distant organs or by direct extension. Their appearance is associated with a poor prognosis and low survival rate. The metastasis rate to this gland by malignant melanoma is 1%, so the approach by cytology is limited, as it shares characteristics with papillary thyroid carcinoma. Case report: a 50-year-old woman, in follow-up for a nodular invasive malignant melanoma in the right frontotemporal region. During follow-up visits, a CT of the neck showed reactive cervical lymph nodes, and an ultrasound revealed a left thyroid lobe nodule that had increased its volume 100% in three months. Discussion: fine needle aspiration revealed epithelial cell papillary formations, irregular nuclei, some exhibiting nuclear inclusions and clefts, being classified as Bethesda VI. A left hemithyroidectomy was performed observing a 1.8 x 1.6 cm whitish mass with irregular margins. Histological analysis showed epithelioid cells, with pleomorphic nuclei and prominent nucleoli. Immunolabeling was positive for SOX-10, S100, HMB45, melan-A and galectin-3 and negative for AE1/AE3, TTF1, thyroglobulin, PAX8 and calcitonin. It was determined that the lesion corresponded to a metastatic melanoma to the thyroid gland. Conclusions: metastases to the thyroid gland are rare but considering melanoma is regarded as one of the simulating neoplasms, it becomes a diagnostic challenge.

Introducción: la patología molecular constituye un eje central de la medicina de precisión en oncología, al integrar hallazgos morfológicos, inmunohistoquímicos y técnicas de biología molecular para identificar biomarcadores en ADN, ARN y proteínas. Estos biomarcadores impactan de manera directa en el diagnóstico, el pronóstico y la selección de terapias personalizadas. Métodos: Se realizó una revisión de la literatura en bases de datos como PubMed, Embase, Scopus, y Google Scholar. Resultados: entre las técnicas más empleadas se incluyen la inmunohistoquímica, la hibridación in situ, la reacción en cadena de la polimerasa y la secuenciación de nueva generación (NGS). Dichas metodologías permiten detectar alteraciones genómicas como variantes puntuales, indeles, variaciones en el número de copias y fusiones, con consecuencias relevantes en la evolución tumoral y la respuesta terapéutica. Discusión: actualmente, se dispone de una amplia gama de biomarcadores en los cuales existe terapía dirigida basado en la alteración molecular puntual. De igual forma, han cobrado importancia los biomarcadores agnósticos que permiten decisiones terapéuticas transversales, independientemente del origen histológico. En paralelo, la evaluación de PD-L1, se ha convertido en una herramienta esencial para seleccionar pacientes candidatos a inmunoterapia. En este escenario de la patología molecular, los biobancos emergen como infraestructuras estratégicas para la medicina de precisión, al garantizar trazabilidad, calidad y estandarización en la gestión de muestras y datos asociados. Conclusión: el Biobanco Nacional de Tumores Terry Fox, en Colombia, representa un ejemplo regional de cómo estas plataformas fortalecen la investigación traslacional y promueven colaboraciones internacionales.