Borderline Z-scores in NIPT: from ambiguity to actionable reporting

Dear Editor,

We read with keen interest Pan et al.’s article on borderline Z-scores in non-invasive prenatal testing (NIPT), which evaluates 83,443 tests and reports that 0.84% of cases fall into an initial “borderline” range. Through library reconstruction and re-sampling, this residual rate is reduced to 0.03%, with an overall abnormality yield of 0.9% [1]. Their study quantifies the uncertainty inherent in borderline calls and importantly demonstrates that re-workflow strategies can meaningfully reduce ambiguous outcomes.

While the work provides a valuable quantitative benchmark, we offer the following refinements to enhance clinical utility, equity, and interpretability of borderline NIPT reporting.

Toward a multifactorial risk assessment model: methodology matters

Pan et al. define “borderline” as Z-scores between + 1.96 and + 3 (or analogous thresholds), yet this static approach overlooks the continuous nature of risk and platform-specific variance in Z-score scaling [1]. A result just above + 1.96 on one instrument may correspond to + 1.8 on another. We propose that every borderline call be supplemented with a post-test risk estimate—complete with uncertainty bounds—derived from observed Z-score, fetal fraction (FF), maternal age, and calibration data.

To operationalize this, we recommend the following modeling framework:

Model type: Bayesian logistic regression or gradient boosting, chosen for their ability to handle non-linear relationships and uncertainty quantification.

Variable selection: use penalized regression (e.g., LASSO) or SHAP-based feature importance to identify robust predictors.

Calibration: apply Platt scaling or isotonic regression to align predicted probabilities with observed outcomes.

Validation: perform temporal and geographic external validation, with subgroup analysis by gestational age, FF strata, and maternal demographics.

Recent work shows that Z-score–based risk transformation often outperforms positive predictive value (PPV) cutoffs in guiding further evaluation [2]. This methodology ensures that risk estimates are both interpretable and generalizable.

Tiered protocol for borderline Z-score management: a flowchart-based framework

Pan et al. also demonstrate that mean FF is lower—and more variable—in samples that remain borderline after repeated workflows [1]. This reinforces that FF is not merely a quality metric but a clinical signal, potentially indicating assay limitations or biological complexity (e.g., mosaicism or placental abnormality) (Fig. 1).

Fig. 1.

Tiered protocol for borderline NIPT results

We propose a tiered protocol, now presented as a flowchart for clarity:

This flowchart outlines a five-step protocol for managing borderline Z-scores in non-invasive prenatal testing (NIPT). It includes initial reporting of Z-score and fetal fraction (FF), eligibility criteria for re-sampling, escalation to diagnostic testing based on FF thresholds, post-negative follow-up for Confined Placental Mosaicism (CPM), and audit metrics to evaluate workflow efficiency and patient burden.

Tiered Protocol for Borderline NIPT results

Step 1: initial report.

Disclose Z-score, FF (overall and per chromosome), and quality flags.

Define borderline as Z-score between + 1.96 and + 3.

Step 2: re-sampling eligibility.

Permit one optimized re-draw if gestational age is ≥ 11 weeks and ≤ 20 weeks.

Ensure standardized pre-analytic handling.

Step 3: escalation criteria.

If FF remains < 4% or borderline persists, recommend direct diagnostic testing (e.g., chorionic villus sampling or amniocentesis).

Step 4: post-negative follow-up.

Even after a resolved low-risk report, borderline calls may indicate Confined Placental Mosaicism (CPM). [3]

Recommend targeted ultrasound, growth surveillance, and placenta-aware obstetric follow-up.

Step 5: audit metrics.

Laboratories should track order → re-sample → reporting times and borderline persistence rates to assess workflow impact on patient burden.

This protocol balances false-positive reduction with minimizing delays and repeat sampling.

Terminology harmonization and reporting standards

We recommend consistent use of “borderline” throughout the manuscript, replacing “gray-zone” for clarity and standardization. Additionally, “CPM” should be introduced with its full term—Confined Placental Mosaicism (CPM)—upon first use.

To address platform heterogeneity and foster transparency, we reiterate our call for a multicenter, de-identified “borderline registry” that aggregates Z-scores, FF metrics, maternal age, workflow interventions, and diagnostic outcomes [2]. Over time, this registry could calibrate shared risk-conversion curves and strengthen external validity.

We also encourage Archives of Gynecology and Obstetrics to require that NIPT studies include supplementary calibration curves, platform definitions, and reproducibility metrics.

Pan et al.’s dataset quantifies the scale of the borderline challenge [1]. Our commentary aims not to critique but to catalyze a shift toward actionable, patient-centric reporting—one that integrates continuous post-test risk, a single retry policy, CPM awareness, and open calibration. If adopted broadly, this framework could transform ambiguous NIPT zones from clinical frustration into structured pathways.

With the journal’s leadership, we hope Archives of Gynecology and Obstetrics will invite authors and laboratories to adopt a minimal standardized framework for borderline reporting, accelerate registry development, and thereby convert statistical uncertainty into better patient care.

Respectfully submitted,

Author contributions

Single author contributed to a study.

Funding

The authors have not disclosed any funding.

Data availability

No datasets were generated or analyzed during the current study.

Declarations

Conflict of interest

The authors declare that there are no competing interests.