Comparison of Kleihauer‐Betke Methodology Performed by Experienced Morphologists and Flow Cytometry in Fetomaternal Haemorrhage Quantification

Fetomaternal haemorrhage (FMH) occurs when there is transplacental transfer of foetal red cells into maternal circulation and can result in alloimmunisation when antigen-negative mothers are exposed. Almost all pregnancies experience a small amount of foetal red cell transfer [1] most commonly during the third trimester and delivery; however, any prenatal trauma, antenatal invasive procedures, or external cephalic version can also result in FMH [2]. Alloantibodies can cross the placenta in subsequent pregnancies, leading to haemolytic disease of the foetus and newborn (HDFN). RhD negative mothers are at greatest risk due to RhD antigen frequency and its high immunogenicity, resulting in rapid IgG alloantibody formation. HDFN due to RhD alloantibodies is largely preventable by RhD immunoglobulin administration to RhD negative mothers through routine antenatal anti-D prophylaxis [3, 4]. According to Australian Guidelines [5], a single 625 IU intramuscular dose of RhD immunoglobulin will protect against an FMH of up to 6 mL of RhD positive foetal red cells, with a further 100 IU RhD immunoglobulin per mL of foetal red cells required for FMH > 6 mL. Accurate detection and quantification of FMH is paramount for RhD immunoglobulin dosing to prevent maternal sensitization against RhD antigen and to investigate foetal welfare. The Australian and New Zealand Society of Blood Transfusion (ANZBT) recommends two methods for estimation of FMH, namely the Kleihauer–Betke test (KB) and flow cytometry (FCM) [6]. KB differentiates foetal from adult red cells by the relative resistance of foetal haemoglobin (HbF) to acid elution compared to adult haemoglobin (HbA) [7] whereby a maternal blood film is exposed to an acid buffer, followed by counterstaining which allows foetal cell numbers to be counted against ‘ghost’ adult cells. FCM detects and enumerates anti-HbF or anti-RhD fluorescent-labelled foetal cells present in maternal blood samples. Inaccuracies in KB results can derive from variation in method, materials, calculations, and inter-operator variability. FCM requires greater resources, operator training and is limited to specialised laboratories. Hence, KB remains the method of choice in many laboratories worldwide with its shorter turnaround time for a time-sensitive clinical issue. The objective of this study is to compare the estimation of FMH in two experienced tertiary laboratories using KB and FCM. Two laboratories with expertise in either KB or FCM FMH testing were identified. Between May 2022 and June 2023, all samples with FMH volume > 2 mL during routine testing were shared with the alternate site. Samples were couriered between sites for analysis to be performed by both methodologies within 96 h from collection. Samples with FMH volume > 2 mL were chosen as these are designated as FMH positive [6]. Statistical analysis was performed using R (version 4.0.2) for Spearman correlation for non-parametric measurements, paired t-test and Bland–Altman plot. p < 0.05 were considered statistically significant. This quality assurance study was approved by New South Wales Health Pathology and determined to be negligible ethical risk. Twenty-one peripheral blood samples were identified as FMH positive during routine screening. Three samples with high maternal HbF were excluded. Three samples were excluded due to logistical transport issues resulting in only one analysis technique performed (two FCM, one KB). In the 15 remaining samples, measured FMH volume ranged from 3.12–141.03 mL by KB and 2.00–110.80 mL by FCM (Table 1). Large FMH bleeds (> 6 mL) were identified in 53.3% (8/15) of cases, and very large FMH bleeds (≥ 15 mL) in 40.0% (6/15) of cases. There was a very high positive correlation between KB and FCM FMH volume measurements, with a Spearman correlation r = 0.977 (p < 0.001) (Figure 1). Exclusion of the three largest FMH volume outliers (Cases 13–15) resulted in a similar high Spearman correlation r = 0.956 (p < 0.001). Bland–Altman plots comparing differences in KB and FCM FMH volumes against the mean FMH volume of both methods showed an absolute positive bias of 4.99 mL towards KB (95% limits of agreement −12.33 to 22.32 mL) (Figure 2A), with mean percentage bias of 19.4% (95% limits of agreement −27.2 to 66.0%) (Figure 2B). Very large FMH volume measurements breach the upper limit of agreement in absolute value, though trend around the mean percentage bias. The number of standard 625 IU RhD immunoglobulin vials required based on KB FMH volume ranged from 1 to 23 vials, and 1 to 18 vials for FCM. There was no difference in the number of RhD immunoglobulin vials required between KB and FCM in 66.7% (10/15), one vial difference in 20% (3/15), and 4–5 vials difference in 13.3% (2/15) of samples (Table 1). Samples with 4–5 vials difference represented the massive FMH volumes (> 100 mL). There has been a trend towards increasing use of FCM in tertiary hospitals with onsite laboratories since the proposal of FCM as the gold standard for quantification of foetal cells and FMH volume, given its precision and high sensitivity. Access to FCM remains a limiting factor, particularly in smaller laboratories where KB remains the method of choice. This study has demonstrated that in a tertiary hospital laboratory, senior well-trained morphology staff are able to achieve highly comparable FMH volumes performed on the same samples using KB to FCM. There is an overall systematic bias of KB to overestimate FMH volumes by a mean absolute value of 4.99 mL in our study. This reduces to a mean absolute bias of 1.80 mL when excluding the measurements with massive FMH volume outliers (> 100 mL), which suggests a better agreement between KB and FCM at lower FMH volumes. In small and large volume FMH bleeds (< 15 mL), the number of RhD immunoglobulin vials required as determined by KB and FCM appears similar (Table 1). In cases with very large FMH bleeds (≥ 15 mL), the FMH volume differences were larger, attributing to the imprecision of quantification by KB at these levels, and leading to greater variation in recommended RhD immunoglobulin dosing. In clinical practice, estimated variations at these very large FMH volumes are less relevant as ongoing repeat FMH quantification would be performed to guide additional RhD immunoglobulin doses. Similar to our study, Urgessa et al. (2014) demonstrated a good correlation (r = 0.897, p < 0.005) between KB and FCM in RhD negative patients in the Ethiopian demographic [8]. Similarly, Pastoret et al. demonstrated good correlation (r = 0.87) using the FMH Quikquant FCM kit [9]. Savithrisowmya et al. demonstrated a good, but lower, correlation (r = 0.75, ICC α = 0.73) [10]. In contrast, Pelikan et al. demonstrated a fair correlation (r = 0.69, weighted κ = 0.40) between manual KB and FCM; however, in relation to the emerging availability and use of digital morphology, there was a good correlation between manual and automatic KB (r = 0.77) with moderate agreement (weighted κ = 0.56) [11]. Further advancements with the use of machine learning by Zhang et al. showing a higher precision between automated KB (ICC α = 0.996) and manual KB (ICC α = 0.790) against FCM highlights the future of digital morphology [12]. Women with high maternal HbF cells, such as in hereditary persistence of foetal haemoglobin and haemoglobinopathies, may pose a challenge to KB FMH quantification given the morphological similarity of maternal HbF cells to foetal red cells after acid elution. The limitations of our study include its small sample size with few cases of large, very large, and massive FMH volumes, as well as exclusion of cases with high maternal HbF cells, reducing its generalizability. There is also a risk of systematic bias secondary to sample handling, noting that almost all cases were analysed by FCM first followed by KB. In conclusion, although FCM-based estimation of FMH remains the analytical method of choice, KB performed with senior and well-trained morphologists is capable of achieving excellent correlation compared to FCM and remains an important test in FMH assessment. Miles Ma: data analysis, preparation of manuscript. Alireza Ardjmand: data collection, data analysis, preparation of manuscript. Nisha Aloysious: preparation of manuscript. Simon Lee: preparation of manuscript. Ioannis Giannoutsos: preparation of manuscript, co-supervisor of study. Giselle Kidson-Gerber: preparation of manuscript, co-supervisor of study. The authors have nothing to report. Reviewed by NSW Health Pathology and determined to be of negligible ethical risk. The authors have nothing to report. The authors declare no conflicts of interest. The data that support the findings of this study are available from the corresponding author upon reasonable request.