Genomic Reporting Practices Across 5 Molecular Disciplines: A Study From the College of American Pathologists Open Access

This study was conducted by members of the CAP’s Molecular Oncology, Genomic Medicine, and Biochemical and Molecular Genetics Committees. The working group consisted of 5 molecular pathologists, 1 hematopathologist, 1 hematopathologist with board certification in molecular pathology, 2 molecular geneticists, 1 statistician, and 1 clinical chemist, as well as 2 CAP staff with expertise in laboratory accreditation, regulatory affairs, and biostatistics. The working group members were divided into subgroups according to their areas of expertise, as follows: oncology (solid tumors and hematologic malignancies), constitutional (germline) genetics, PGx, and prenatal cfDNA. The subgroup members met periodically via conference calls, and meetings were held among all working group members.

The groups reviewed regulatory requirements (42 CFR § 493.1291), accreditation criteria (CAP checklists related to molecular pathology and general laboratory practices [Molecular Pathology, All Common, and Laboratory General]), professional guidelines for reporting,^1,2,4–6  examples of clinical genomic reports, and published literature on reporting of molecular test results.^9,10,12–16 

Patient–de-identified genomic reports were obtained from 31 clinical laboratories/companies in each of 5 molecular disciplines. Some laboratories submitted multiple reports. Most of the reports, except for PGx, were based on next-generation sequencing (NGS). Molecular microbiology and cytogenomics reports were not included in this study. Reports were reviewed independently by 2 working group members and a CAP staff liaison and were scored for the presence or absence of applicable reporting elements (Table 1; Supplemental Table 1, see supplemental digital content containing 6 tables at https://meridian.allenpress.com/aplm in the July 2025 table of contents), of which 23 were elements required by CAP and/or CLIA to be included in a clinical report (Table 1). Report content was informed by CAP/CLIA requirements, published literature on molecular reporting, and group consensus. If present, the element’s location on the report was recorded, which included 5 options: header, footer, methods, interpretation, and results. The element location was a “select all” format, since some elements appeared in multiple locations (Supplemental Tables 2 through 6). Some of the elements were only specifically pertinent to 1 or to a subset of molecular disciplines (eg, tumor percentage for oncology reports and star alleles for PGx reports), which is why the number of reviewed elements ranged from 27 to 44.

The groups used “N/A” for an element if it was not applicable for a discipline and recorded as “absent” data elements that were relevant but not included in a report. Missing elements and reporting elements with more than 4 N/A scores were excluded from the data analysis. Based on this criterion, none of the CAP/CLIA-required reporting elements were excluded. Owing to a limited number of reports containing results on gene fusions, large insertions and deletions, copy number variants, and DNA/RNA microdeletions/microduplications (eg, FLT3 internal tandem duplications), these elements were also excluded from the analysis. Familiarity by working group members with individual assay report characteristics was used to assist with content representation in some instances.

For each of the 5 disciplines, frequency distributions were calculated to summarize the presence of the required reporting elements in addition to the element location(s) on the report. The analysis of reporting element location was performed only when the element was present on at least 5 reports. Logistic regression models with a Firth correction were used to test element-present rates between report types. All summaries and analyses were performed with SAS 9.4 (SAS Institute, Cary, North Carolina), and a significance level of .05 was used for the testing.

Eighty-one clinical reports (73 containing abnormal findings and 8 negative) obtained from 31 different laboratories/companies were reviewed. These included 24 germline reports, 39 oncology reports (hematologic malignancies, n = 21; solid tumors, n = 18), 11 PGx reports, and 7 prenatal cfDNA reports. Academic and commercial laboratories contributed 80 reports, and 1 report was obtained from a pathology reference center for the federal government. Laboratories were geographically dispersed throughout the United States. Twelve reports were excluded from analysis for various reasons, including negative reports, non–NGS-based assay (with the exception of PGx), discontinued assays, or updated assay versions. A total of 69 reports were analyzed: 17 germline, 20 hematologic malignancies, 17 solid tumors, 9 PGx, and 6 prenatal cfDNA.

The 17 germline reports were derived from diagnostic exome assays (6), diagnostic targeted panels (5), diagnostic single-gene assay (1), and carrier tests (5). All oncology assays were targeted genomic panels. Five PGx reports were obtained from targeted NGS panels and 4 were derived from single-gene polymerase chain reaction–based genotyping assays. The latter were included because it is a commonly used method for PGx testing. All prenatal cfDNA assays comprised genome-wide analysis of chromosomal aneuploidies and microdeletions/microduplications of selected chromosomal regions.

Overall, 71% (49 of 69) of reports included all CLIA-required elements (Table 2). Patient demographics (100%, 69 of 69), test name (100%, 69 of 69), variant interpretation (97%, 67 of 69), technology used for testing (94%, 65 of 69), laboratory information (94%, 65 of 69), laboratory director/designee signature (93%, 64 of 69), analyte-specific reagents statement (93%, 64 of 69), assay limitations (87%, 60 of 69), and technical details of the assay (83%, 57 of 69) were listed in most of the reports (Figure; Table 1; Supplemental Tables 2 through 6). Of those, the only report element that reached a statistically significant difference in frequency rate across the 5 molecular disciplines was assay technical details (P = .02) (Table 1). Overall, 14% (10 of 69) of reports contained all CAP accreditation–required elements (Tables 1 and 3).

Some of the CAP-required elements are not pertinent to all molecular disciplines. For instance, star allele and metabolizer phenotype are exclusively applicable to PGx reports. Consequently, these data were only present in the PGx reports (Table 1; Supplemental Table 5). Conversely, variant tiers/classification is not an element currently applicable to PGx and prenatal cfDNA aneuploidy testing reports; thus, it was not listed in any of the reports of this discipline reviewed in the study. Likewise, gene names/symbols and variants detected were listed in all germline, somatic, and PGx reports, but were not applicable to the prenatal cfDNA reports included in the study (Figure; Table 1). Discipline-specific reporting elements were only analyzed for the disciplines they are applicable to.

Of the basic reporting elements required for all tests, patient demographics was the only element consistently displayed in a specific report section (header) across all disciplines. Laboratory information was listed in the header and/or footer of the reports. Test names were generally presented in the report header across different disciplines. Of those that reported a clinical indication, it was listed in the header of all germline (16), hematologic malignancy (16), and prenatal cfDNA (6) reports. Two of the 17 solid tumor somatic reports displayed this information in the methods and results sections. Clinical indication was typically not included in the PGx reports, likely because PGx tests usually target specific markers to identify patients who may be at risk for altered metabolism of drugs (Supplemental Table 1).

Most reports—100% (15 of 15) solid tumors, 100% (5 of 5) prenatal cfDNA, 90% (17 of 19) hematologic malignancies, 63% (5 of 8) PGx, and 59% (10 of 17) germline—listed the laboratory director’s signature in the footer, although if reported, this element was also displayed in methods, results, and interpretation sections. When reported (64 of 69, 93%), the analyte-specific reagents disclaimer was placed in the footer of most reports but was also listed in methods and interpretation sections of a few reports from all disciplines (Supplemental Tables 2 through 6).

Variants were reported from the complementary DNA sequence (cDNA) in 100% (17 of 17), 95% (19 of 20), 94% (16 of 17), and 67% (6 of 9) of germline, hematologic malignancy, solid tumor, and PGx reports, respectively (P = .04). In 94% (16 of 17) of germline, 90% (18 of 20) of hematologic malignancy, 82% (14 of 17) of solid tumor, and 44% (4 of 9) of PGx reports, sequence variations were described in relation to a reference sequence (ie, reference transcript ID was provided) (P = .02). However, genome build was listed in only 62% of the reports that included this data element (39 of 63) (P < .001). All germline and somatic reports provided the predicted protein change(s) and position(s) (p.) of detected variants (Figure; Table 1). Variant reporting overall followed the Human Genome Variation Society nomenclature, although there were variations in the use of single-letter versus 3-letter amino acid notations. Variant allele frequency was included in 95% (19 of 20) and 82% (14 of 17) of hematologic malignancy and solid tumor reports, respectively. Zygosity and disease inheritance pattern, which are data elements primarily applicable to germline reports, were listed in 94% (16 of 17) and 100% (17 of 17) of those reports, respectively (Supplemental Table 1).

Variant classification was provided in 94% (16 of 17) of germline and 78% (29 of 37) of somatic reports (P = .35) (Figure; Table 1). A tiered classification system^1,2  was used by 94% (16 of 17) of germline, 82% (14 of 17) of solid tumor, and 75% (15 of 20) of hematologic malignancy reports for variant reporting (Supplemental Table 1). The remaining laboratories either reported variants based on pathogenicity or listed the variants without any classification assignments. Variant interpretations were included in all but 2 of the 20 hematologic malignancy reports (P = .59) (Table 1). Of those, 82% (49 of 60) interpreted variants from the specific disease/disorder in question: 100% (17 of 17) germline, 100% (6 of 6) prenatal cfDNA, 77% (13 of 17) solid tumors, and 65% (13 of 20) hematologic malignancies. Of note, PGx reports traditionally report on predicted drug response phenotype rather than focusing on the specific underlying disorder for which the drug will be prescribed. Clinicopathologic correlations (ie, correlation of clinical findings with histopathologic findings and with the results of genomic tests) were present in 94% (16 of 17) of solid tumor, 60% (12 of 20) of hematologic malignancy, 17% (1 of 6) of prenatal cfDNA, and 12% (2 of 17) of germline reports (P < .001) (Figure; Table 1).

The most consistent clinical recommendation element was genetic counseling, provided in 100% of germline (17 of 17) and prenatal cfDNA (6 of 6) reports. Clinical trial information was present in 35% (6 of 17) of solid tumor and 30% (6 of 20) of hematologic malignancy reports (Supplemental Table 1).

Information related to transplants, pregnancies, comorbidities, among others—pertinent to the interpretation of genomics findings—was included in all 6 prenatal cfDNA reports (6 of 6, 100%) and in 75% (12 of 16) of germline reports (Supplemental Table 1).

Pertinent negatives, corresponding to variants in clinically relevant genetic regions not detected in the sample, were included in about half of oncology reports (53%, 9 of 17 solid tumors; and 40%, 8 of 20 hematologic malignancies). Pertinent negatives were not listed in the germline, PGx, and prenatal cfDNA reports, likely owing to the nature of these reports (ie, comprehensive exome, single gene assessment, carrier testing, and targeted genotyping) (Supplemental Table 1). For PGx, laboratories routinely interrogate the same set of variants and when negative, report as *1/*1 and normal metabolizer rather than stating the negative positions.

Despite variations in report formats across laboratories and disciplines, the placement of interpretive information was overall consistent, often redundant across different report sections, and present in sections where the information would be expected to be found in germline, hematologic malignancy, and solid tumor reports (Supplemental Tables 2 through 6). In these disciplines, gene names, variants detected, variant nomenclature, variant allele frequency, zygosity, variant classifications, and variant interpretation, when provided, were listed in the results and/or interpretation sections. In PGx reports, gene names and star alleles were placed in the results, interpretation, and/or methods sections, and zygosity and metabolizer phenotype were reported in the results and/or interpretation sections (Supplemental Table 5). Prenatal cfDNA reports placed test results summary/interpretation in the results and interpretation sections (Supplemental Table 6).

Within and across disciplines, testing methodology reporting varied from a detailed description of procedures to more basic statements. Assay targets (ie, a list of covered genes) were provided in all 20 hematologic malignancy reports and 77% (13 of 17) of solid tumor reports (P = .04) (Table 1). The germline, PGx, and prenatal cfDNA reports did not contain this information, since the former is by definition comprehensive (eg, exome sequencing) and the latter 2 are often limited to a single gene assessment.

Assay limit of detection (LOD) was listed in 100% (20 of 20) of hematologic malignancy and in 77% (13 of 17) of solid tumor reports (P = .04) (Figure; Table 1). Given that variants of germline origin are generally present at 50% (heterozygous) or 100% (homozygous) allele frequency, LOD information was not included in the germline and PGx reports. Since prenatal cfDNA is a screening assay and relies on statistical analyses of fetal cfDNA in the maternal blood, reporting LOD information would not be applicable. Tumor cell percentage was reported in only 29% (5 of 17) of solid tumor reports.

Of the technical elements recommended, but not required to be present in sequencing reports of different disciplines, coverage depth at variant level and minimum mean assay target coverage depth were listed in only a small proportion of reports. Coverage depth was provided in 29% (5 of 17) of germline and in approximately 40% (8 of 20 and 7 of 17, respectively) of both solid tumor and hematologic malignancy somatic reports. Minimum mean assay target coverage depth was present in 82% (14 of 17) of germline, 75% (15 of 20) of hematologic malignancy, and 53% (9 of 17) of solid tumor reports (Supplemental Table 1). Coverage information was not applicable to prenatal cfDNA reports, which comprised aneuploidy testing, and to PGx reports that were derived from polymerase chain reaction–based genotyping assays.

A smaller proportion of somatic reports (70% [26 of 37], both hematologic malignancies [15 of 20] and solid tumors [11 of 17]), provided detailed technical information about the assays, compared with germline and PGx reports, in which technical details were present in all the reports ([17 of 17] and [9 of 9], respectively), although this difference was not statistically significant (P = .02) (Figure; Table 1).

If reported, assay technology, technical details, and assay limitations were placed in the methods section of all germline, solid tumor, and PGx reports, and in most of the hematologic malignancy reports. One prenatal cfDNA report included assay limitations in both header and methods sections of the report, with this element being reported in the methods section for the remaining prenatal cfDNA reports (Supplemental Tables 2 through 6).

LOD was listed in the methods section of all hematologic malignancy (20 of 20, 100%) and most solid tumor (13 of 17, 77%) reports that included this element. If reported, the minimum mean assay target coverage depth was also placed in the methods section of all hematologic malignancy reports (15 of 15, 100%) and in all but 1 germline (13 of 14, 93%) and 1 solid tumor (8 of 9, 89%) report, where this element was listed in the interpretation section in both cases. The placement of variant coverage depth, when provided, varied within and across disciplines, with locations including interpretation, results, and methods sections (Supplemental Tables 2 through 6).

Pertinent clinical information related to transplants, pregnancies, and comorbidities, among others, was listed in the methods section of all germline reports that included this element (12 of 16, 75%). All prenatal cfDNA reports (6 of 6, 100%) included clinical information in the header (5) or footer (1) section (Supplemental Tables 2 through 6).

Clinical recommendations, such as family member testing (13 of 14, 93%), genetic counseling (17 of 17, 100%), gene-disease associations (17 of 17, 100%), population-based (5 of 15, 33%) and individual predictive (5 of 17, 29%) values, and carrier status and incidental findings (10 of 14, 71%), were placed in either the methods, interpretation, or results sections of all reports that included these elements (Supplemental Table 2). All reports (8 of 17, 47%) that included procedures pertaining to variant reclassification listed this element in their methods section.

Clinicopathologic correlations, variant confirmation status, variant detection rates, and the specific criteria used for variant classification (eg, American College of Medical Genetics and Genomics PVS1, PM1, BP1) were each listed in fewer than 5 reports, and therefore not analyzed for location.

Overall, clinical trial information was included in 35% (6 of 17) of solid tumor and 30% (6 of 20) of hematologic malignancy reports. This element was included in the results and interpretation sections of all but 1 solid tumor report, where this element was listed in the methods section. Pertinent negatives, when provided (53% [9 of 17] of solid tumor and 40% [8 of 20] of hematologic malignancy reports), were most commonly reported in the results section of both hematologic malignancy and solid tumor somatic reports (75% [6 of 8] and 89% [8 of 9], respectively). In contrast, this information was listed in the interpretation section of 2 hematologic malignancy reports and 1 solid tumor report. Tumor mutation burden was reported in 47% (8 of 17) of solid tumor reports. When available, this element was reported in the results (8) and/or interpretation (4) sections. False-positive/negative rates and detection rates did not reach the minimum number of reports for location analysis.

The predicted phenotype information (eg, metabolizer phenotype) was listed in the results (78%, 7 of 9) and/or interpretation (44%, 4 of 9) sections of the 9 PGx reports.

Risk assessment for negative and positive screening test results was included in the results (80%, 4 of 5), interpretation (80%, 4 of 5), and methods (20%, 1 of 5) sections of the reports that included this element (5 of 6, 83%).

Clinical recommendations for confirmatory testing and genetic counseling appeared in the methods section of 83% (5 of 6) and 67% (4 of 6) of reports, respectively. These reporting elements were also listed in the results and/or interpretation sections.

All but 1 of the 6 prenatal cfDNA reports included in this study followed a modular/distributive testing model, whereby different components of the test were performed by more than 1 CLIA-certified laboratory. Information about the laboratories to which testing was referred to as part of the modular testing approach was provided in the footer (3 of 5, 60%), header (1 of 5, 20%), and methods (1 of 5, 20%) sections of the reports.

Our study shows that clinical genomic reports are overall compliant with regulatory requirements, although some reporting elements are less consistently reported. Of the 69 reports, 62 (90%) included 6 of the 7 CLIA-required elements. However, 71% (49) contained all CLIA-required elements and 10 of the 69 reports (14.5%) had all the required CAP accreditation elements. Despite variability in the content and presentation of the information, there is a trend toward consistent placement of reporting elements across laboratories and disciplines.

Variations in reporting styles are not unexpected given the inherent differences in clinical disciplines, practice settings, and the lack of specific regulatory requirements for standardization of clinical report formatting. According to CLIA (42 CFR §493.1291), laboratories should assess the needs of laboratory users when determining the format, style, and language of molecular genetic test reports. These are expected to vary based on the indication of testing, practice setting, and level of genomics literacy of the intended users of these genetic tests. In addition, laboratories differ in their information technology infrastructure, resources, preferences, constraints, and customer characteristics, which would make an overarching standardization of genomics report formats logistically challenging.

Several studies have provided recommendations for the design and content of genetic test reports.^1–4,16,17  However, owing to the lack of conclusive experimental data (eg, randomized trials), it is uncertain if any specific standardized report format improves communication and clinician understanding of genomic test results. Not surprisingly, in studies where user feedback was obtained regarding the proposed layout and content of reports, there was diversity of opinions and preferences.

Nonetheless, evidence suggests that data presentation plays a relevant role in communication of information between laboratories and clinicians. Consistency in document layout and text formatting can improve comprehension, reduce the risk of misinterpretation, and prevent reported information from being overlooked. In fact, consistency in the way data are positioned in a document has been proposed to be more relevant for efficient communication than the actual location where the information is displayed.¹⁸  This is corroborated by findings from a study by Powsner et al⁷  in which misunderstanding of pathology report results considerably increased following stylistic changes, which was in part attributed to the lack of familiarity of physicians with the updated report format (“report disorientation”). Although the readability of genomic test reports was not specifically evaluated in this study, differences in format and location of information on reports from different laboratories, particularly within molecular disciplines, may pose challenges to physicians to make clinically important management decisions. Since treating physicians may receive molecular reports from various internal and external laboratories, the array of report styles and formats could compound report disorientation.

This study is limited by the fact that although a reasonable number of representative examples of genomic reports were obtained, representing 5 molecular disciplines, the sampling was restricted to the reports accessible to and reviewed by the study team. There may be inherent biases, and the evaluated reports cannot be assumed to illustrate the general practices of all laboratories within the United States. Moreover, not all report examples included all types of variants. There were sparse data (<5 reports) for several variant categories such as fusion events, structural changes, and copy number variants, which were excluded from analysis. Not all the reports were obtained from NGS assays, and certain reporting elements evaluated in this study are NGS-specific, although most of the report elements assessed in this study are pertinent to molecular diagnostics assays in general.

While the adoption of synoptic reporting for genomic assays with inclusion of discipline-specific reporting elements, as previously proposed,^4,12,19  could assist with report harmonization, completeness, and interoperability, we acknowledge that the development of a standardized framework for reporting genomics data in a synoptic or structured format, as well as the format and content of the information to be captured, should be evidence based, outcome oriented, and assessed in terms of the ability to benefit clinical decision-making. Understanding genomic reporting practices is a necessary first step to guide the future development of discipline-specific reporting elements for synoptic genomic reporting and molecular data integration into pathology reports.

The authors thank the laboratories that provided reports for this study, and Gwendolyn McMillian, PhD, for assisting with the review of PGx reports.