Journal Club – CAD

Broome CM, Cunningham JM, Mullins M, et al. Increased risk of thrombotic events in cold agglutinin disease: a 10-year retrospective analysis. Res Pract Thromb Haemost. 2020;4(4):628–635.

Clinical Question

Among adults with cold agglutinin disease (CAD), is the risk of a first thrombotic event higher than in matched patients without CAD after adjustment for comorbidity and prior thrombotic risk?

Background

CAD is a rare complement-mediated autoimmune hemolytic anemia driven by IgM autoantibodies and classical pathway activation. Hemolysis is predominantly C3-mediated and extravascular, although some patients also have terminal complement-mediated intravascular hemolysis. Thrombosis is well recognized in other hemolytic disorders, but the burden of thrombosis in CAD has been less clear. Prior reports were limited to small series, leaving uncertainty about whether CAD itself contributes to thrombotic risk beyond age and comorbidities. This study used a large US claims-linked clinical database to address that question.

Guidelines

There are no guidelines regarding the use of thromboprophylaxis in patients with CAD. The 2020 International Concensus Meeting highlights that more active hemolysis and prior splenectomy are can be associated with an increased risk of venous thromboembolism (VTE); however the decision to use thromboprophylaxis should take into account an individual patient’s disease-specific risk factors as well as general VTE risk factors¹

Study Design

• Design: Retrospective matched cohort study.
• Data source: Optum deidentified Integrated Claims–Clinical database, linking electronic medical record and adjudicated claims data from approximately 55 million US patients.
• Study period: 2006–2016.
• CAD identification: Clinical notes were searched for CAD-related terms. Patients with CAD terms on at least 3 separate dates were classified as CAD cases (n=517). Patients with terms on 1 or 2 dates underwent independent hematologist review, yielding 91 additional cases; validation of this approach showed 95% agreement.
• Matching: Each CAD case was matched to up to 10 controls without CAD based on age (±3 years), sex, race, region, duration of activity on Optum, and season/year of entry into the database.
• Statistical analysis: Cox regression was used to estimate time to first thrombotic event, adjusting for age, sex, race, region, active time in Optum, prior thrombosis, major comorbid conditions and treatment exposures, matched cluster, and Charlson Comorbidity Index. Adjusted absolute risk differences at 1 and 5 years were calculated using bootstrap confidence intervals.

Population

• Cohort size: 608 patients with CAD matched to 5,873 patients without CAD, with a mean of 9.7 controls per case.
• Baseline demographics: Approximately 70% were age ≥65 years, 63% were female, 85% were White, and the largest regional representation was from the Midwest.
• Residual imbalance despite matching: The CAD cohort had higher baseline rates of prior thrombosis (18.1% vs 11.3%), malignant cancer (27.0% vs 9.9%), organ failure (17.3% vs 8.1%), chemotherapy exposure (13.2% vs 3.2%), and anticoagulant use (28.0% vs 15.7%).
• Follow-up and mortality: Mean follow-up was shorter in CAD than controls (25.0 vs 35.3 months), and mortality was higher (19.6% vs 4.7%).
• Exclusions: Age younger than 25 years at index, less than 1 year of pre-index enrollment, or less than 1 month of post-index follow-up.

Outcomes

• Primary outcome: First thrombotic event after the index date, identified from inpatient or outpatient ICD-9/10 codes.
• Event categories: Venous, arterial, and cerebral thrombotic events were analyzed separately.
• Sensitivity analyses:
  • Presumed primary CAD, excluding patients with diagnoses associated with secondary CAD such as lymphoma, myeloma, CLL, Waldenström macroglobulinemia, and selected infections.
  • VTE-only analyses using 2 previously published Sanfilippo algorithms².
  • Inpatient-only thrombotic events with a 6-month look-back exclusion to reduce duplicate counting.
  • Stratification by Charlson Comorbidity Index to assess risk across comorbidity burden
• Key Results:
  • Overall thrombotic risk: Thrombotic events occurred in 29.6% of patients with CAD compared with 17.6% of matched controls, corresponding to incidence rates of 14.2 versus 6.0 per 100 person-years. The cumulative incidence curves separated early and remained significantly different throughout follow-up.
• By event type:

• Absolute risk difference: The adjusted absolute excess risk was 11.9 thrombotic events per 100 patients at both 1 and 5 years, suggesting that the risk emerged early and persisted over time.
• Sensitivity analyses: The association between CAD and thrombosis remained consistent across multiple alternative analyses:
  • Presumed primary CAD: adjusted HR 1.80 (95% CI, 1.46–2.22).
  • VTE, Sanfilippo algorithm 1: adjusted HR 2.95 (95% CI, 2.28–3.82).
  • VTE, Sanfilippo algorithm 2: adjusted HR 3.10 (95% CI, 2.24–4.30).
  • Inpatient-only thrombotic events: adjusted HR 1.87 (95% CI, 1.40–2.51).
• Stratified by comorbidity burden: The relative risk was greatest among patients with the lowest Charlson Comorbidity Index (CCI).
  • CCI 0: 22.1% vs 10.6%; adjusted HR 2.44 (95% CI, 1.70–3.52).
  • CCI 1–2: 27.9% vs 20.0%; adjusted HR 2.05 (95% CI, 1.56–2.68).
  • CCI ≥3: 40.6% vs 33.2%; adjusted HR 1.57 (95% CI, 1.14–2.16).

Commentary

This study makes a compelling case that thrombosis is part of the clinical phenotype of CAD, not incidental to it. In this matched cohort, CAD was associated with nearly double the overall risk of thrombosis and nearly triple the risk of venous thrombosis. The association held up across several sensitivity analyses, including restriction to presumed primary CAD, alternative VTE definitions, and stratification by comorbidity burden. This does not establish causation, but it does prompt heightened clinical vigilance for thromboembolic complications.

The biology is plausible. CAD combines chronic hemolysis with ongoing classical complement activation, both of which can promote thrombosis through nitric oxide depletion, endothelial dysfunction, inflammation, tissue factor expression, and procoagulant microparticles. The observation that the relative risk was highest in patients with CCI=0 is especially interesting, because it suggests that CAD itself may contribute to thrombosis rather than simply tracking with medical complexity. Several features strengthen the paper: it was the largest CAD cohort available at the time, it used a national linked claims-clinical dataset, case identification was strengthened by note review and hematologist adjudication, and the sensitivity analyses were directionally consistent. The signal is also consistent with a previously published population-based Danish cohort³.

Still, there are real limits to what this study can tell us. It is retrospective and claims-based, so thrombotic events were identified by ICD codes rather than by imaging review or formal adjudication. The dataset could not capture the hemolytic variables, including hemoglobin, LDH, complement activity, cold agglutinin titer, thermal amplitude, or the timing and intensity of CAD-directed therapy. Primary and secondary CAD could not be cleanly separated in the main analysis, and baseline differences in malignancy, prior thrombosis, chemotherapy exposure, and anticoagulant use leave room for residual confounding. Shorter follow-up in the CAD cohort, likely related to higher mortality, also raises the possibility of informative censoring. Whether complement inhibition can reduce thrombotic events in CAD, as it does in PNH, remains an open question⁴,⁵. In practice, this study supports recognizing CAD as a prothrombotic condition and maintaining a low threshold for thromboprophylaxis in high-risk clinical settings.

References

1. Jäger U, Barcellini W, Broome CM, Gertz MA, Hill A, Hill QA, Jilma B, Kuter DJ, Michel M, Montillo M, Röth A, Zeerleder SS, Berentsen S. Diagnosis and treatment of autoimmune hemolytic anemia in adults: Recommendations from the First International Consensus Meeting. Blood Rev. 2020 May;41:100648. doi: 10.1016/j.blre.2019.100648. Epub 2019 Dec 5. PMID: 31839434. ↩︎
2. Sanfilippo KM, Wang TF, Gage BF, Liu W, Carson KR. Improving accuracy of International Classification of Diseases codes for venous thromboembolism in administrative data. Thrombosis research. 2015 Apr 1;135(4):616-20. ↩︎
3. Bylsma LC, Gulbech Ording A, Rosenthal A, et al. Occurrence, thromboembolic risk, and mortality in Danish patients with cold agglutinin disease. Blood Advances. 2019/10/22;3(20)doi:10.1182/bloodadvances.2019000476 ↩︎
4. Röth A, Berentsen S, Barcellini W, et al. Long-term efficacy and safety of continued complement C1s inhibition with sutimlimab in cold agglutinin disease: CADENZA study Part B. eClinicalMedicine. 2024/08/01;74doi:10.1016/j.eclinm.2024.102733 ↩︎
5. Hillmen P, Muus P, Dührsen U, et al. Effect of the complement inhibitor eculizumab on thromboembolism in patients with paroxysmal nocturnal hemoglobinuria. Blood. 2007/12/01;110(12)doi:10.1182/blood-2007-06-095646 ↩︎

Rishabh Singh, MD, is a Hematology–Oncology Fellow at Indiana University. He earned his medical degree from Topiwala National Medical College in Mumbai, India, and completed his Internal Medicine residency at the University of Illinois College of Medicine in Peoria, Illinois. His clinical and research interests focus on thrombosis, with a specific focus on anticoagulation stewardship, thrombotic microangiopathies, and the role of neutrophils in thrombosis.