What is the approach to risk stratification in patients?

Risk Stratification Approach in Clinical Practice

Risk stratification is fundamentally the categorization of patients into risk strata based on demographic variables, comorbidities, and disease-specific factors to predict the probability of developing adverse events, with the primary goal of improving health outcomes through early detection and appropriate intervention before fatal or nonfatal events occur. 1

Core Principles of Risk Stratification

Risk stratification serves to prognosticate rather than determine whether to initiate treatment, which should be based on established diagnostic criteria for symptomatic disease 1. The process involves identifying patients at relatively high risk for major events, though it must be recognized that the majority of adverse events actually occur in low- to intermediate-risk populations, while the highest-risk subgroups constitute only a small proportion of total events 1.

Two Categories of Risk Stratifiers

Modifiable Risk Factors

• Include conditions that can be detected and changed through avoidance (smoking) or intervention (hypertension, hypercholesterolemia) 1
• Data from general populations show these changes reduce cardiovascular event incidence 1
• Critical caveat: Some conditions show paradoxical associations in specific populations (e.g., higher BMI associated with better outcomes in dialysis patients, opposite to general population) 1

Fixed Risk Factors

• Include age, gender, genetically determined conditions, and biomarkers 1
• These cannot be changed but identify patients with damaged or high-risk organ systems 1
• Examples include troponins and specific genetic variants 1

Disease-Specific Risk Stratification Algorithms

Acute Coronary Syndromes

For acute presentations (unstable angina, NSTEMI), immediate risk stratification using TIMI or GRACE scores is mandatory 2:

• TIMI Score calculation: Assign 1 point for each of 7 variables (age ≥65 years, ≥3 CAD risk factors, known CAD, aspirin use in prior 7 days, severe angina, ST-segment deviation ≥0.5mm, elevated cardiac biomarkers) 2
• Score interpretation: 0-2 = low risk, 3-4 = intermediate risk, 5-7 = high risk 2

High-risk features requiring immediate catheterization: refractory angina, hemodynamic instability, sustained ventricular tachycardia/fibrillation, acute heart failure with pulmonary edema, or recurrent angina with ST-segment changes ≥2mm 2

Intermediate-risk features: Plan catheterization within 24-72 hours for elevated troponin, dynamic ST/T-wave changes, diabetes with positive biomarkers, reduced renal function, depressed LVEF, early post-MI angina, or prior PCI/CABG 2

Perioperative Cardiac Risk

Use the Revised Cardiac Risk Index (RCRI) assigning 1 point for each of 6 predictors: ischemic heart disease, congestive heart failure, cerebrovascular disease, high-risk surgery, insulin-dependent diabetes, and preoperative kidney dysfunction 3

Risk classification by RCRI score:

• 0 factors = low risk
• 1 factor = low to moderate risk
• 2 factors = moderate risk
• ≥3 factors = high risk 3

Management algorithm:

• RCRI 0-1 with good functional capacity (≥4 METs): Proceed without additional cardiac testing 3
• RCRI ≥2 or poor functional capacity: Consider additional cardiac evaluation including stress imaging, coronary CTA, or BNP measurement 3

Hematologic Malignancies

For multiple myeloma, risk stratification combines cytogenetics, disease burden, and patient factors 1:

High-risk features (median OS ~3 years): FISH showing del 17p, t(14;16), t(14;20), or high-risk GEP signature 1

Intermediate-risk features (median OS 4-5 years): FISH t(4;14), cytogenetic del 13, hypodiploidy, or PCLI ≥3% 1

Standard-risk features (median OS 8-10 years): All others including t(11;14), t(6;14) 1

For pre B-cell ALL, stratification is determined sequentially 4:

1. First: Philadelphia chromosome (BCR-ABL1) status
2. Second: Age-based categorization
3. Third: Cytogenetics (hypodiploidy <44 chromosomes = very poor prognosis with 5-year EFS 13-24%; high hyperdiploidy 51-65 chromosomes = favorable with 5-year EFS 49-50%) 4
4. Fourth: WBC count (≥30×10⁹/L for B-cell = high risk; ≥100×10⁹/L for T-cell = high risk) 4
5. Fifth: Minimal residual disease (MRD) after induction (EOI MRD ≥0.01% = higher risk) 4

Critical point: MRD status after induction therapy is the strongest independent prognostic factor, superseding traditional clinical variables 4

Practical Implementation Considerations

Timing of Risk Stratification

• At diagnosis: Establish baseline risk to guide initial treatment intensity 1
• Post-induction/intervention: Reassess based on response markers (e.g., MRD, troponin trends) 2, 4
• At relapse: Re-stratify as risk profiles may change 1

Common Pitfalls to Avoid

Do not rely solely on age as a risk factor; focus on specific cardiac, pulmonary, and renal disease status 3

Recognize population-specific paradoxes: Risk factors validated in general populations may show reversed associations in specific disease states (dialysis patients, advanced malignancies) 1

Avoid premature treatment mandates based solely on genetic abnormalities until definitive evidence shows specific therapies overcome poor-risk features 1

Understand that risk is a continuum, not a dichotomy, and most adverse events occur in low-to-intermediate risk groups despite lower individual risk 1

Quality of Risk Stratification Tools

The ideal risk stratification tool should:

• Identify most patients who will experience adverse events while excluding those who will not 1
• Lead to interventions that improve survival to a greater extent than in similar patients with normal results 1
• Be easily applicable in clinical practice with readily identifiable subpopulations 5

Important limitation: Many tools provide prognostic information on both disease-specific and non-disease-specific mortality, which can complicate interpretation 1

Integration with Novel Therapies

Emerging evidence suggests novel agents (bortezomib, lenalidomide) may overcome some poor-risk features in myeloma, but current data does not yet mandate specific therapy selection based on cytogenetic abnormalities alone 1. Clinical trials should continue evaluating whether specific treatments can overcome traditional poor-risk factors 1.

AI/ML applications for point-of-care risk stratification are under development but face implementation challenges including need for external validation, clinical workflow integration, and addressing bias/fairness concerns 1. These approaches are likely near-to-mid-term solutions rather than immediate clinical tools 1.