Risk prediction models for colorectal cancer: Evaluating the discrimination due to added biomarkers

The Use of Blood-Based Biomarkers in the Prediction of Colorectal Neoplasia at the Time of Primary Screening Colonoscopy Among Average-Risk Patients: A Systematic Literature Review

BACKGROUND/OBJECTIVES

Risk prediction models (RPMs) for colorectal cancer (CRC) could facilitate risk-based screening. Models incorporating biomarkers may improve the utility of current RPMs. We performed a systematic review of studies reporting RPMs for CRC that evaluated the impact of blood-based biomarkers on clinical outcome prediction at the time of screening colonoscopy in average-risk populations.

METHODS

We conducted a search of MEDLINE, Web of Science, and PubMed databases from inception through April 2024. Studies that developed or validated a model to predict risk of CRC or its precursors were included. Studies were limited to those including patients undergoing average-risk CRC screening.

RESULTS

Sixteen studies published between 2015 and 2024 were included. Outcomes included CRC (16 studies) and high-risk adenomas (1 study). Using a complete blood count was the most common biomarker and was able to achieve an AUC of 0.82 and a specificity of 0.88. Other blood-based biomarkers included were various serum proteins/metabolites/enzymes, plasma metabolites, insulin-related factors, and anemia markers. The highest-performing model, with an AUC of 0.99, involved the use of a plasma metabolite panel.

CONCLUSIONS

The evidence base of RPMs for CRC screening is expanding and incorporating biomarkers, which remain a prominent aspect of model discovery. Most RPMs included a lack of internal/external validation or discussion as to how the model could be implemented clinically. As biomarkers improve the discriminatory potential of RPMs, more research is needed for the evaluation and implementation of RPMs within existing CRC screening frameworks.

Plasma proteomic and polygenic profiling improve risk stratification and personalized screening for colorectal cancer

This study aims to identify colorectal cancer (CRC)-related proteomic profiles and develop a prediction model for CRC onset by integrating proteomic profiles with genetic and non-genetic factors (QCancer-15) to improve the risk stratification and estimate of personalized initial screening age. Here, using a two-stage strategy, we prioritize 15 protein biomarkers as predictors to construct a protein risk score (ProS). The risk prediction model integrating proteomic profiles with polygenic risk score (PRS) and QCancer-15 risk score (QCancer-S) shows improved performance (C-statistic: 0.79 vs. 0.71, P = 4.94E-03 in training cohort; 0.75 vs 0.69, P = 5.49E-04 in validation cohort) and net benefit than QCancer-S alone. The combined model markedly stratifies the risk of CRC onset. Participants with high ProS, PRS, or combined risk score are proposed to start screening at age 46, 41, or before 40 years old. In this work, the integration of blood proteomics with PRS and QCancer-15 demonstrates improved performance for risk stratification and clinical implication for the derivation of risk-adapted starting ages of CRC screening, which may contribute to the decision-making process for CRC screening.

Metabolomic signatures of inflammation and metabolic dysregulation in relation to colorectal cancer risk

BACKGROUND

Inflammation and metabolic dysregulation are associated with increased risk of colorectal cancer (CRC); the underlying mechanisms are not fully understood. We characterized metabolomic signatures of inflammation and metabolic dysregulation and evaluated the association of the signatures and individual metabolites with CRC risk.

METHODS

Among 684 incident CRC cases and 684 age-matched controls in the Nurses' Health Study (n = 818 women) and Health Professionals Follow-up Study (n = 550 men), we applied reduced rank and elastic net regression to 277 metabolites for markers of inflammation (C-reactive protein, interleukin 6, tumor necrosis factor receptor superfamily member 1B, and growth differentiation factor 15) or metabolic dysregulation (body mass index, waist circumference, C-peptide, and adiponectin) to derive metabolomic signatures. We evaluated the association of the signatures and individual metabolites with CRC using multivariable conditional logistic regression. All statistical tests were 2-sided.

RESULTS

We derived a signature of 100 metabolites that explained 24% of variation in markers of inflammation and a signature of 73 metabolites that explained 27% of variation in markers of metabolic dysregulation. Among men, both signatures were associated with CRC (odds ratio [OR] = 1.34, 95% confidence interval [CI] = 1.07 to 1.68 per 1-standard deviation increase, inflammation; OR = 1.25, 95% CI = 1.00 to 1.55 metabolic dysregulation); neither signature was associated with CRC in women. A total of 11 metabolites were individually associated with CRC and biomarkers of inflammation or metabolic dysregulation among either men or women.

CONCLUSION

We derived metabolomic signatures and identified individual metabolites associated with inflammation, metabolic dysregulation, and CRC, highlighting several metabolites as promising candidates involved in the inflammatory and metabolic dysregulation pathways for CRC incidence.

Machine Learning-Based Predictive Modeling of Diabetic Nephropathy in Type 2 Diabetes Using Integrated Biomarkers: A Single-Center Retrospective Study

Purpose

Diabetic nephropathy (DN), a major complication of diabetes mellitus, significantly impacts global health. Identifying individuals at risk of developing DN is crucial for early intervention and improving patient outcomes. This study aims to develop and validate a machine learning-based predictive model using integrated biomarkers.

Methods

A cross-sectional analysis was conducted on a baseline dataset involving 2184 participants without DN, categorized based on their development of DN over a follow-up period of 36 months: DN (n=1270) and Non-DN (n=914). Various demographic and clinical parameters were analyzed. The findings were validated using an independent dataset comprising 468 participants, with 273 developing DN and 195 remaining as Non-DN over the follow-up period. Machine learning algorithms, alongside traditional descriptive statistics and logistic regression were used for statistical analyses.

Results

Elevated levels of serum creatinine, urea, and reduced eGFR, alongside an increased prevalence of retinopathy and peripheral neuropathy, were prominently observed in those who developed DN. Validation on the independent dataset further confirmed the model's robustness and consistency. The SVM model demonstrated superior performance in the training set (AUC=0.79, F1-score=0.74) and testing set (AUC=0.83, F1-score=0.82), outperforming other models. Significant predictors of DN included serum creatinine, eGFR, presence of diabetic retinopathy, and peripheral neuropathy.

Conclusion

Integrating machine learning algorithms with clinical and biomarker data at baseline offers a promising avenue for identifying individuals at risk of developing diabetic nephropathy in type 2 diabetes patients over a 36-month period.

Endoscopic Screening and Risk of Colorectal Cancer according to Type 2 Diabetes Status

Current recommendations for colorectal cancer screening have not accounted for type 2 diabetes (T2D) status. It remains unknown whether the colorectal cancer-preventive benefit of endoscopic screening and the recommended age for screening initiation differ by T2D. Among 166,307 women (Nurses' Health Study I and II, 1988-2017) and 42,875 men (Health Professionals Follow-up Study, 1988-2016), endoscopic screening and T2D diagnosis were biennially updated. We calculated endoscopic screening-associated hazard ratios (HR) and absolute risk reductions (ARR) for colorectal cancer incidence and mortality according to T2D, and age-specific colorectal cancer incidence according to T2D. During a median of 26 years of follow-up, we documented 3,457 colorectal cancer cases and 1,129 colorectal cancer deaths. Endoscopic screening was associated with a similar HR of colorectal cancer incidence in the T2D and non-T2D groups (P-multiplicative interaction = 0.57). In contrast, the endoscopic screening-associated ARR for colorectal cancer incidence was higher in the T2D group (2.36%; 95% CI, 1.55%-3.13%) than in the non-T2D group (1.73%; 95% CI, 1.29%-2.16%; P-additive interaction = 0.01). Individuals without T2D attained a 10-year cumulative risk of 0.35% at the benchmark age of 45 years, whereas those with T2D reached this threshold risk level at the age of 36 years. Similar results were observed for colorectal cancer mortality. In conclusion, the absolute benefit of endoscopic screening for colorectal cancer prevention may be substantially higher for individuals with T2D compared with those without T2D. Although T2D is comparatively rare prior to the fifth decade of life, the rising incidence of young-onset T2D and heightened colorectal cancer risk associated with T2D support the consideration of earlier endoscopic screening in individuals with T2D.

PREVENTION RELEVANCE

The endoscopic screening-associated ARRs for colorectal cancer incidence and mortality were higher for individuals with T2D than those without T2D. Endoscopic screening confers a greater benefit for colorectal cancer prevention among T2D individuals, who may also benefit from an earlier screening than the current recommendation.

Pro/Antioxidant State as a Potential Biomarker of Schizophrenia

To allow better diagnosis and management of psychiatric illnesses, the use of easily accessible biomarkers are proposed. Therefore, recognition of some diseases by a set of related pathogenesis biomarkers is a promising approach. The study aims to assess the usefulness of examining oxidative stress (OS) in schizophrenia as a potential biomarker of illness using the commonly used data mining decision tree method. The study group was comprised of 147 participants: 98 patients with schizophrenia (SZ group), and the control group (n = 49; HC). The patients with schizophrenia were divided into two groups: first-episode schizophrenia (n = 49; FS) and chronic schizophrenia (n = 49; CS). The assessment included the following biomarkers in sera of patients: catalase (CAT), glutathione peroxidase (GPx), superoxide dismutase-1 (SOD-1), glutathione reductase (GR), reduced glutathione (GSH), total antioxidant capacity (TAC), ferric reducing ability of plasma (FRAP), advanced glycation end products (AGEs), advanced oxidation protein products (AOPP), dityrosine (DITYR), kynurenine (KYN), N-formylkynurenine (NFK), tryptophan (TRY), total oxidant status (TOS), nitric oxide (NO) and total protein. Maximum accuracy (89.36%) for distinguishing SZ from HC was attained with TOS and GPx (cut-off points: 392.70 and 15.33). For differentiating between FS and CS, the most promising were KYN, AOPP, TAC and NO (100%; cut-off points: 721.20, 0.55, 64.76 and 2.59). To distinguish FS from HC, maximum accuracy was found for GSH and TOS (100%; cut-off points: 859.96 and 0.31), and in order to distinguish CS from HC, the most promising were GSH and TOS (100%; cut-off points: 0.26 and 343.28). Using redox biomarkers would be the most promising approach for discriminating patients with schizophrenia from healthy individuals and, in the future, could be used as an add-on marker to diagnose and/or respond to treatment.