Optimizing TREC- and KREC-based newborn screening: Risk-stratified algorithms significantly reduce referrals.

BACKGROUND

Newborn screening (NBS) quantifying T-cell receptor excision circles with or without kappa-deleting recombination excision circles (TREC, KREC) enables early detection of severe T- and/or B-cell lymphopenia. However, both markers have limited specificity, often resulting in unnecessary referrals. We compared the effects of alternative risk-stratification strategies to reduce referrals while avoiding missed cases and unnecessary delays.

METHODS

We modeled the effects of multiple TREC and KREC-based NBS algorithms using a comprehensive real-life dataset from the first 6 years of the Swiss NBS program. Stratification approaches included adjustment of cut-offs and integration of clinical data such as gestational age (GA), postmenstrual age (PMA), inpatient vs. outpatient status, family history, and maternal immunosuppression.

RESULTS

Lowering cut-offs alone reduced abnormal results by 42% for TREC and 64% for KREC. The most efficient TREC algorithm, based on exact TREC levels and post-menstrual age in preterm infants, reduced referrals by 61% (p < .0001) but missed nearly half of non-SCID T-cell lymphopenias and delayed referral in 2/3 of outpatients who eventually required immunological evaluation. Integration of family history and clinical signs mitigated delays in some cases. For KREC, combining information on gestational age, maternal immunosuppression, and inpatient status enabled a >10-fold reduction in referrals (p < .0001) while still identifying all confirmed agammaglobulinemia cases except two with λ5 deficiency.

CONCLUSION

Risk-stratified, multistep NBS algorithms incorporating readily available clinical data can substantially reduce unnecessary referrals while preserving detection of target conditions. For KREC, simple algorithmic adjustments allow marked improvement in specificity with minimal diagnostic loss.