Ninety-three specimens from 85 patients were processed in the validation study. These specimens comprised 55 chorionic villi, 20 umbilical cord, five formalin-fixed paraffin-embedded tissue (FFPET), four fetal membrane, skin and cultured cell samples and one whole-blood sample. The tissue samples were microscopically dissected and 5–30 mg of chorionic villus was disassociated in a mixture of 1.25% trypsin in PBS/Versene, 0.2 mg/ml collagenase (C1889; Sigma, St. Louis, MO, USA) and 20% fetal calf serum in Ham’s F10 medium. Similar quantities of other tissues were disassociated in collagenase only. An aliquot from each specimen was placed in a micro-centrifuge tube and stored at −20°C for up to 3 weeks. This approach was used to ensure the efficient batching of samples while maintaining the 28-day turnaround time mandated by the National Pathology Accreditation Council (NPAAC). The remaining specimens were used to produce slides for subsequent confirmation by fluorescence in situ hybridization (FISH). Slides were frozen at −20°C until they were required for use. Explant and/or cell suspension cultures were established on all the specimens for routine G-banded karyotyping.

DNA extraction from defrosted cell suspensions was performed using a column-based procedure (ZR Genomic DNA™ Tissue MiniPrep; Zymo Research, Orange, CA, USA) according to the manufacturer’s instructions, and DNA was recovered in 25 μl elution buffer. The DNA from FFPET was extracted using a standard phenol-chloroform procedure after the specimen was dewaxed using xylene. The eluted DNA concentration and purity were assessed using a NanoDrop1000 spectrophotometer (Thermo Scientific, Thermo Electric North America LLC, Waltham, MA, USA).

KaryoLite™ BoBs was performed according to the manufacturer’s instructions (PerkinElmer Life Sciences Wallac, Turku, Finland) and the beads were analyzed using a Luminex 100 platform equipped with a BMD FIDIA system and MLX booster™ software (Biomedical Diagnostics, Marne La Vallee, France). Two male and two female controls were included in each session for normalization. The results were analyzed and converted to numerical values using BoBsoft 1.0 software (PerkinElmer Life Sciences Wallac). Mosaicism and maternal cell contamination yielded ambiguous results, with ratios that differed from those expected. Suspected mosaicism was confirmed using FISH. Standard karyotyping was performed on all the viable tissues and the results were compared with the BoBs data. Abnormal results that were not detected by G-banding were confirmed by FISH, using locus-specific probes.

Specimens from an additional 95 patients were prepared as outlined above, with the exception that a small amount of cell suspension was cultured as a back-up only. This was not processed further unless required for the confirmation of dosage changes.

ezANOVA (McCausland Centre, Columbia, SC, USA) and SigmaXL (version 2.1; Sigma XL Inc., Toronto, Canada) programs were used for data analysis. A non-parametric Kruskal-Wallis test was also performed (raw data not presented).

The validation study yielded 63 normal results with approximately equal male/female ratios. Since 18 tissues failed to grow in culture, the results were verified by FISH alone. Thirty abnormal results were obtained, including two samples in which results were obtained from >1 tissue type. This gave an adjusted abnormality frequency of 32.9%, which is skewed as a number of tissues were selected on the basis of an abnormal G-banding result in order to validate the BoBs technique.

The majority of abnormalities were identified as common aneuploidies associated with fetal loss, namely trisomies 13, 16, 18, 21 and 22 (Table II). Approximately 30% of the abnormal results represented an abnormal ploidy level. BoBs was unable to unequivocally identify these and thus required FISH verification. Among the discrepant results, an unbalanced translocation, described by G-banding as 46, XY, der(6)t(6;10)(p25.1;q25.2), was identified to be male with additional 10qter, and a marker from a case with a 92,XXX-X,+mar karyotype was shown to be 9qter material. A double trisomy (male +14, +22) was identified by BoBs and verified using FISH in 100% of ≥50 cells that were examined, while G-banding on cultured cells identified only the trisomy 14 (47,XY,+14); the additional chromosome 22 was not observed in any of the metaphase cells examined. A gain of 20qter was identified by BoBs and verified by FISH; however, this gain was likely to be below the resolution of the light microscope as G-banding identified it as a normal female.

The prospective study involved an additional 95 patients and yielded 71 normal results with a slightly higher female/male ratio and an overall adjusted abnormality frequency of 27%. As experienced in the validation study, the commonly encountered trisomies associated with pregnancy loss dominated the range of abnormal results. A trisomy 18 case also had an apparently balanced translocation between chromosomes 5 and 10; however, this was not identified by BoBs. Three aneuploidies (XO, trisomy 13 and monosomy 21) revealed slightly ambiguous results due to high correlation coefficient values (CV; 8.5–10.9), which may have occurred as a result of the quality of DNA extracted from the tissues. We also identified two samples that showed an XYY signal pattern of <2% on direct FISH due to extensive maternal contamination, which were later confirmed as entirely XYY by FISH probes applied to the paraffin-embedded tissues. A G-banded karyotype showing an add (4p) was characterized as a gain of 14p. The remaining abnormalities are shown in Table II.

Alkuraya et al(12) reported a high incidence of cryptic subtelomeric abnormalities in miscarriage products. Therefore, we anticipated a greater number of abnormal occurrences in telomeric regions compared with other chromosomal regions analyzed by the BoBs, even when FISH was unable to verify them. Notably, statistical analysis of all the abnormal calls by the BoBs software showed that there was no significant difference between subtelomeric calls and other chromosomal regions. Eight chromosomes were revealed to account for ~60% of the calls; however, this group also included the chromosomes most often encountered in pregnancy loss, namely chromosomes 16, 18, 20, 21 and 22 (raw data not shown).

Following the validation and prospective studies, a retrospective analysis of our earlier G-banding data for the years 2006–2010 (inclusive) was performed (unpublished data). A total of 1,573 samples were processed by the laboratory. The overall culture failure rate was 20% and the abnormality frequency was 22.4%. FFPET samples that were received for ploidy studies were not included in this analysis.

We hypothesized that BoBs would correctly identify all the aneuploids and unbalanced karyotypes that were detectable by light microscopy. We predicted that five cases with abnormal karyotypes would yield non-concordant results using BoBs technology (Table III). Case 1 was a balanced translocation, case 2 was a mosaic with two cell lines, one of which was a translocation and case 3 was an interstitial deletion. These three cases represent <0.2% of the total cases processed. The remaining two cases involved abnormalities that BoBs was expected to detect. In both cases BoBs would also be expected to identify the origin of the marker chromosome by detecting increased dosage of centromeric material, provided the cell line containing the marker was above 30% in the uncultured sample. Bearing in mind, almost 300 cases did not yield any results due to tissue culture failure and assuming the BoBs failure and abnormality pick-up rates were correct at the time, we hypothesized that the use of BoBs may have identified an additional 60 abnormal results.