Subjectivity in chromosome band-level estimation: a multicenter study.

PURPOSE: Chromosome band level is the primary quality indicator for G-banded metaphase chromosome analysis. Although current professional guidelines address the minimum necessary band level for constitutional studies, there is no study documenting the comparative performance of different band-level estimation methods. METHODS: This study compared 5 band-level estimation methods (Stallard, Vancouver, Welborn, United Kingdom External Quality Assurance Scheme, and Ford) in a multicenter study in which 82 readers from 7 different clinical cytogenetics laboratories evaluated the same 10 karyotypes (5 from amniotic fluid and 5 from peripheral blood) by each method. RESULTS: There was a 94% correlation between the five band-level estimation methods. The Welborn method yielded significantly lower scores for amniotic fluid karyotypes (P < 0.01) but not for peripheral blood karyotypes (P = 0.75). The distribution of scores obtained from different readers suggests a high level of subjectivity in chromosome band-level assessment. The variation in band-level estimation did not correlate with reader experience or study center, except for readers from one laboratory, for which the distribution of scores was significantly lower (P < 0.01). CONCLUSION: The results from this study suggest that the consistent use of one method is more important than the actual method employed for monitoring karyotype quality.

Use Reference Bands to Accurately Estimate ISCN Band Levels 400, 550, and 850.

In their 2002 Guidelines for chromosome analysis of peripheral blood, the American College of Medical Genetics states that "The 550-band stage should be the goal of all constitutional studies..." The College of American Pathologists requires that the average case be analyzed at the 400-band level of resolution for routine work, and that the 550-band level be achieved in appropriate blood samples. The challenge is how to identify the 400, 550, and 850-band levels confidently and consistently. In this study, our objectives were to develop simple and reliable criteria to estimate band level, and to evaluate our laboratory's performance with respect to those criteria. Using the ISCN(1995) ideogram as a reference, candidate bands were selected for the three band levels: 400, 550 and 850. A pilot and two follow-up studies were conducted and a set of candidate bands were validated against the Vancouver method of evaluating band level so that band level scores were similar using either method. The final set of reference bands were the presence of 9q32 and 20q13.2 for the 400-band level; 5q33.2 and 10q22.2 for the 550-band level; and 3p26.1, 18q22.3 and 20q13.32 for the 850-band level. Cell selection improved after each technologist was provided a composite image of chromosomes with reference bands highlighted. The band level criteria presented here involve no band counting, appear to be objective, can help to improve quality and consistency among technologists, and can ensure compliance with regulatory agencies.