Longitudinal evaluation of external quality assessment results for CA 15-3, CA 19-9, and CA 125

Background

Tumor markers are established laboratory tools that help to diagnose, estimate prognosis, and monitor the course of cancer. For meaningful decision-making in patient care, it is essential that methods and analytical platforms demonstrate high sensitivity, specificity, precision, and comparability. Regular participation at external quality assessment (EQA) schemes is mandatory for laboratories. Here, a longitudinal evaluation of EQA data was performed to assess the performance of tumor marker assays over time.

Methods

Longitudinal data of the cancer antigens (CA) 15-3 (n = 5,492), CA 19-9 (n = 6,802), and CA 125 (n = 5,362) from 14 INSTAND EQAs conducted between 2019 and 2023 were evaluated. A median of 197, 244 and 191 laboratories participated at the EQAs for CA 15-3, CA 19-9 and CA 125, respectively. Data evaluation encompasses intra- and inter-manufacturer specific variations over time, assay precision, and adherence to the EQA limits of ±24% for CA 15-3, ±27% for CA 19-9 and ±36% for CA 125.

Results

The study showed median manufacturer-dependent differences of up to 107% for CA 15-3, 99% for CA 125, and even 549% for CA 19-9 between the highest and the lowest methods over the studied period. Regarding the normalized median of all methods, the values of the most deviant methods were 0.42 for CA 15-3, 7.61 for CA 19-9, and 1.82 for CA 125. Intra-manufacturer variability was generally low, with median coefficients of variation (CV) below 10%. As the methods were evaluated according to method-specific consensus values, most participants passed the EQAs within the acceptance criteria. When the criteria were consistently set at 24%, the central 90% of participants passed the EQAs in 78.6%-100% for CA 15-3 (with exception of AX), 89.3%-100% for CA 125, and 64.3%-100% for CA 19-9.

Conclusion

While intra-method precision of most analytical platforms is acceptable for all three tumor markers, considerable inter-method variability was observed over the whole studied period demonstrating the necessity for better standardization and harmonization of the methods, development of international reference materials, and comprehensive commutability studies with patient samples.

Method Comparison and Clinical Performance of Breast Cancer Tumor Markers on Novel Multiplex Immunoassay and Automatized LOCI Technology Platforms

Tumor marker determinations are valuable tools for the guidance of breast cancer patients during the course of disease. They are assessed on diverse analytical platforms that may be associated with differences according to the methods applied and the clinical performance. To investigate the method dependency and clinical significance of breast cancer protein tumor markers, CEA, CA 15-3, CA 125, CA 19-9 and AFP were measured in a total of 154 biobanked samples from 77 patients with breast cancer, 10 with DCIS, 31 with benign breast diseases and 36 healthy controls using a Millipore multiplex biomarker panel (MP) and an automized version of the routinely used Vista LOCI technology. The markers were compared between methods and investigated for diagnostic performance. CEA, CA 15-3 and AFP showed good correlations between both platforms with correlation coefficients of R = 0.85, 0.85 and 0.92, respectively, in all samples, but similarly also in the various subgroups. CA 125 and CA 19-9 showed only moderate correlations (R = 0.71 and 0.56, respectively). Absolute values were significantly higher for CEA, CA 15-3, CA 125 and AFP in the Vista LOCI as compared with the MP method and vice versa for CA 19-9. The diagnostic performance for discrimination of breast cancer from healthy controls was similar for both methods with AUCs in ROC curves for CEA (MP 0.81, 95% CI 0.72-0.91; LOCI 0.81; 95% CI 0.72-0.91) and CA-15-3 (MP 0.75, 95% CI 0.65-0.86; LOCI 0.67, 95% CI 0.54-0.79). Similar results were obtained for the comparison of breast cancer with benign breast diseases regarding CEA (AUC MP 0.62, 95% CI 0.51-0.73; LOCI 0.64, 95% CI 0.53-0.74) and CA-15-3 (MP 0.70, 95% CI 0.6-0.81; LOCI 0.66, 95% CI 0.54-0.77). Both platforms show moderate to good method comparability for tumor markers with similar clinical performance. However, absolute levels in individual patients should be interpreted with care.

Clinical performance of LOCI™-based tumor marker assays for tumor markers CA 15-3, CA 125, CEA, CA 19-9 and AFP in gynecological cancers

Evidence is sparse regarding the clinical performance of luminescent oxygen channeling immunoassays-based tumor marker assays in gynecological cancer. Analyzing serum samples of 336 patients with Dimension™Vista1500, we investigated the diagnostic power of carbohydrate antigen 15-3, carbohydrate antigen 125, carcinoembryonic antigen, carbohydrate antigen 19-9, and alpha-fetoprotein in patients suffering from different types of gynecological cancer and precancerous gynecological diseases and compared findings to appropriate control groups. The cohort comprised 177 female patients with gynecological cancers (73 breast, 22 cervical, 16 endometrial, 17 vulva, and 49 ovarian cancers), 26 patients with precancerous gynecological diseases (11 vulva, 4 cervical, and 10 breast), 109 patients with benign gynecological diseases, and 24 healthy controls. Discriminative power was assessed by areas under the curve in receiver operating characteristic curves, and sensitivities were determined at a fixed specificity of 95%. Levels of biomarkers in healthy controls were in the expected ranges and a discriminative power between gynecological cancers and healthy controls was observed for several tumor markers. Established tumor type-associated markers were elevated in specific gynecological cancers and benign controls as well as within precancerous gynecological diseases and healthy control group. In ovarian cancer, carbohydrate antigen 125 and carbohydrate antigen 15-3 were significantly elevated compared to the respective benign diseases. Carbohydrate antigen 125 was the most conclusive marker (area under the curve = 0.86% and 77.6% sensitivity at 95% specificity). In breast cancer, carcinoembryonic antigen and carbohydrate antigen 15-3 were significantly higher than in the respective benign diseases. Carcinoembryonic antigen achieved the most conclusive area under the curve (0.65) with 31.5% sensitivity at 95% specificity. None of the investigated markers was found to be of value in discriminating benign and malignant cervical diseases. Carcinoembryonic antigen and alpha-fetoprotein distinguished precancerous breast and vulva diseases from healthy controls. These findings show that luminescent oxygen channeling immunoassays-based tumor marker assays provide reliable results in routine diagnostics.

Interassay Comparison of the Tumor Markers CA125, CA15.3, and CA27.29

BACKGROUND

Cancer antigens 125, 27.29, and 15-3 (CA125, CA27.29, and CA15-3) are markers of ovarian and breast cancer. Comparing tumor marker results across methods is challenging because of the lack of harmonization. Documenting comparability of results is important.

METHODS

Siemens Advia Centaur CA125 and CA27.29 assays were compared to their corresponding Beckman Coulter DxI CA125 and CA15-3 assays. The interassay bias was determined and the manufacturer-recommended reference intervals were evaluated.

RESULTS

The DxI CA125 assay demonstrated an overall positive 29% bias relative to the Centaur CA125 assay. The DxI CA15-3 assay demonstrated an overall negative 65% bias relative to the Centaur CA27.29 assay. For patients with multiple comparisons during the study period, the trend of results over time was similar across both sets of assays. Implementing the manufacturer-recommended reference interval for the DxI CA125 assay increased the abnormal flagging rate by 4.5%. In contrast, implementing the manufacturer-recommended reference interval for the DxI CA15-3 assay decreased the abnormal flagging rate by 13.0%.

CONCLUSIONS

The overall trends for the majority of patients were similar. Therefore, despite the overall biases, transitioning tumor marker assays should not affect clinical interpretation of results.

CA-125, but not galectin-3, complements CA 19-9 for discriminating ductal adenocarcinoma versus non-malignant pancreatic diseases

BACKGROUND/OBJECTIVES

CA 19-9 is the gold standard biomarker of pancreatic adenocarcinoma despite several weaknesses in particular a high rate of false positives or negatives. CA-125 corresponding to MUC16 and galectin-3, a lectin able to interact with mucin-associated carbohydrates, are tumor-associated proteins. We investigated whether combined measurement of CA 19-9, galectin-3 and CA-125 may help to better discriminate pancreatic adenocarcinoma versus non-malignant pancreatic diseases.

METHODS

We evaluated by immunohistochemistry the expression of MUC4, MUC16 (CA-125) and galectin-3 in 31 pancreatic adenocarcinomas. We measured CA 19-9, CA-125 and Gal-3 in the serum from patients with pancreatic benign diseases (n = 58) or adenocarcinoma (n = 44). Clinical performance of the 3 biomarkers for cancer diagnosis and prognosis was analyzed.

RESULTS

By immunohistochemistry, MUC16 and Gal-3 were expressed in 74% and 84% of adenocarcinomas versus 0% and 3.2% in peri-tumoral regions, respectively. At the serum level, CA 19-9 and CA125 were significantly higher in patients with pancreatic adenocarcinoma whereas Gal-3 levels did not differ. The performance of CA 19-9 for cancer detection was higher than those of CA-125 or Gal-3 by ROC analysis. However, CA-125 offers the highest specificity for malignancy (81%) because of an absence of false positives among type 2 diabetic patients. Cancer deaths assessed 6 or 12 months after diagnosis varied according to the initial CA-125 level (p < 0.006).

CONCLUSION

Gal-3 is not an interesting biomarker for pancreatic adenocarcinoma detection. CA 19-9 alone exhibits the best performance but measuring CA-125 provides complementary information in terms of diagnosis and prognosis.

Evaluation of Beckman Coulter DxI 800 immunoassay system using clinically oriented performance goals

OBJECTIVES

We evaluated the analytical performance of 24 immunoassays using the Beckman Coulter DxI 800 immunoassay systems at Mayo Clinic, Rochester, MN for trueness, precision, detection limits, linearity, and consistency (across instruments and reagent lots).

METHODS

Clinically oriented performance goals were defined using the following methods: trueness-published desirable accuracy limits, precision-published desirable biologic variation; detection limits - 0.1 percentile of patient test values, linearity - 50% of total error, and consistency-percentage test values crossing key decision points. Local data were collected for precision, linearity, and consistency. Data were provided by Beckman Coulter, Inc. for trueness and detection limits.

RESULTS

All evaluated assays except total thyroxine were within the proposed goals for trueness. Most of the assays met the proposed goals for precision (86% of intra-assay results and 75% of inter-assay results). Five assays had more than 15% of the test results below the minimum detection limits. Carcinoembryonic antigen, total thyroxine and free triiodothyronine exceeded the proposed goals of ±6.3%, ±5% and ±5.7% for dilution linearity. All evaluated assays were within the proposed goals for instrument consistency. Lot-to-lot consistency results for cortisol, ferritin and total thyroxine exceeded the proposed goals of 3.3%, 11.4% and 7% at one medical decision level, while vitamin B12 exceeded the proposed goals of 5.2% and 3.8% at two decision levels.

CONCLUSIONS

The Beckman Coulter DxI 800 immunoassay system meets most of these proposed goals, even though these clinically focused performance goals represent relatively stringent limits.