For many diseases, particularly cancer, no single biomarker test offers acceptable clinical sensitivity and specificity. To obtain more clinically actionable results, clinical laboratories have turned to combinations of biomarkers to better reveal the truth of a patient’s condition.

Often referred to as multianalyte assays with algorithmic analyses (MAAAs), these tests combine results from two or more biochemical or molecular markers, along with patient demographics and clinical information, into an algorithm to generate diagnostic, prognostic, or predictive information for a disease. In cases where single biomarker tests lack acceptable clinical sensitivity and specificity, MAAAs can improve or refine disease detection through individualized risk assessment.

MAAAs fall into two categories based on whether they use proprietary or nonproprietary algorithms. Examples of MAAAs with proprietary algorithms include GenPath’s 4Kscore and Vermillion’s OVERA assays for prostate and ovarian cancer risk assessment, respectively. MAAAs with proprietary algorithms may or may not be approved by the Food and Drug Administration (FDA) and are performed exclusively in a single laboratory. MAAAs with nonproprietary algorithms are FDA-approved and available from commercial laboratories. Examples of MAAAs with nonproprietary algorithms include the Prostate Health Index (phi) from Beckman Coulter and the Risk of Ovarian Malignancy Algorithm (ROMA) for prostate and ovarian cancer risk assessment, respectively.

In recent years interest has grown in seeing MAAAs for various diseases developed and clinically adopted, including for cancers, acute kidney injury, sepsis, and preeclampsia. However, MAAAs have been a standard tool in obstetric care for more than 30 years to screen and to identify pregnancies that may have an increased risk of birth defects such as trisomy 21 (Down syndrome) and trisomy 18 (Edwards syndrome). Here we discuss current MAAAs as they relate to women’s health.

Ovarian Cancer

Women with adnexal masses suspected of ovarian malignancy benefit from referral to gynecologic oncologists for surgical intervention, as triaging to these subspecialists has been shown to improve patient outcomes. While most masses are benign, 13%-21% are malignant. Differentiating between benign and malignant disease and appropriately triaging women with malignant masses poses a challenge for physicians. Historically, physicians used cancer antigen-125 (CA-125) to evaluate ovarian cancer. Due to this marker’s inadequate diagnostic sensitivity and specificity, laboratories and diagnostic manufacturers introduced several algorithms to estimate the risk of malignancy in women presenting with an adnexal mass, including the Risk of Malignancy Index (RMI), OVA1, OVERA, and ROMA (Table 1). RMI incorporates CA-125, ultrasound features, and menopausal status for preoperative risk stratification, while the others use a combination of serum biomarkers as detailed below.

In 2009 FDA cleared OVA1 to assess the probability that an adnexal mass is benign or malignant prior to a planned surgery. OVA1 incorporates the serum concentrations of CA-125, transferrin, transthyretin, apolipoprotein A-1 (ApoA-1), and β2-microglobulin in conjunction with menopausal status into a proprietary algorithm that generates a risk score ranging from 0-10. OVA1 showed a sensitivity of 92%, compared to CA-125 with a sensitivity of 74%, for detecting malignancy in women presenting with an adnexal mass (1). However, OVA1 possesses less than optimal specificity of 54%, resulting in a high false-positive rate and a positive predictive value of only 31% (1).

In 2016 FDA cleared OVERA, the second generation of OVA1. OVERA substitutes two of the five OVA1 biomarkers (transthyretin and β2-microglobulin) with human epididymis protein-4 (HE-4) and follicle stimulating hormone (FSH) and does not require menopausal status. This second-generation assay has improved specificity (69%) while maintaining similar sensitivity (2). Both of these tests use predefined cutoffs to categorize women into low or elevated risk of finding ovarian cancer at surgery.

Fujirebio Diagnostics introduced ROMA in 2010 as a nonproprietary MAAA that uses serum concentrations of CA-125 and HE-4 along with menopausal status to classify women with an adnexal mass as low- or high-risk of being found with an ovarian malignancy at surgery. ROMA cutoffs for low- and high-risk stratification optimize specificity over sensitivity. At 75% specificity, the sensitivity of ROMA is 77% and 92% in premenopausal and postmenopausal women, respectively (3). At the same sensitivity, ROMA has been shown to be more specific than CA-125. 

A prospective study that compared the performance of OVA1 and ROMA in 146 patients with surgically confirmed malignancies found that OVA1 was more sensitive than ROMA, 97% versus 87% (4). However, ROMA was more specific than OVA1, 83% versus 55% (4). The American College of Obstetricians and Gynecologists (ACOG) Practice Bulletin (Number 174, November 2016) outlining the clinical guidelines for evaluating and managing adnexal masses considers the use of both tests as Level B recommendations (based on limited or inconsistent scientific evidence). It recommends that clinicians refer premenopausal or postmenopausal women with an elevated OVA1 or ROMA score to a gynecologic oncologist (5).

Breast Cancer

Physicians use MAAA testing as a prognostic tool in breast cancer and to tailor patients’ treatment based on the unique biology of their cancer. Historically, immunohistochemistry (IHC) testing, along with measurements such as tumor size, tumor grade, and lymph node status, was the standard in guiding breast cancer therapy. Testing for estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2) guides the use of specific targeted therapies.

Although IHC discriminates among breast cancer subsets and helps identify appropriate therapies, more information on the cancer’s molecular biology can be gained through gene expression analysis, which may have a greater ability to predict outcomes and tailor therapies. Several gene expression profiling tests are now available for use in patients diagnosed with breast cancer. These were designed to supplement existing techniques (Table 2). 

With these tests, expression levels of various genes, alone or in combination with information about tumor size and lymph node status, are applied to an algorithm to gain both prognostic and predictive information about a patient’s cancer. The algorithm generates a numerical value that classifies a tumor as low-, intermediate-, or high-risk—or in the case of the MammaPrint assay, patients are simply categorized as having a good- or a poor-prognosis signature. Classifications correlate with the probability of cancer recurrence within 5-10 years.

Gene expression profiling testing also helps predict whether a patient will benefit from adjuvant therapy in addition to standard hormone therapy. Patients at low-risk are likely to do well with hormone therapy alone and can forgo adjuvant chemotherapy. For example, in the Oncotype DX breast cancer test for women diagnosed with ER-positive and node-negative invasive breast cancer, results from the gene expression profile are combined into a recurrence score (RS) reported as a number between 0 and 100. A low RS means a patient likely will receive minimal benefit from the addition of chemotherapy since the cancer has a lower chance of recurrence. On the other hand, a patient with a high RS may have significant benefits from adjuvant chemotherapy since there is an increased risk of cancer recurrence. In patients with RS in the intermediate zone (11-25), the benefits of adjuvant chemotherapy have been unclear until recently. A study of 6,711 women with hormone-receptor-positive, HER2-negative, axillary-node-negative breast cancer with RS between 11-25 who were randomly assigned to two treatment groups, endocrine therapy alone or endocrine therapy plus chemotherapy, found that in patients with an intermediate RS there was no benefit of adjuvant chemotherapy as both groups had similar overall survival rates of 93.9% and 93.8%, respectively (Sparano et al 2018). 

Some breast cancer MAAAs, such as Prosigna, differentiate among molecular subtypes of breast cancer. This assay is based on a 50-gene classifier algorithm along with clinical information about tumor size and lymph node status and is intended for use in postmenopausal women with hormone receptor-positive breast cancer.

The test categorizes breast cancers into one of four molecular subtypes: Luminal A, Luminal B, HER2-enriched, or basal-like, each with differing prognoses. For example, Luminal A tumors typically have low proliferation and high hormone receptor expression. These tumors are associated with a low-risk score and good prognosis. Conversely, Luminal B tumors characteristically have high proliferation rates, making these patients candidates for adjuvant chemotherapy (6).

Breast cancer gene expression profile assays differ significantly in gene sets, analytical platforms, and the patient populations used in their development and validation. The genes included are related to cell proliferation, cancer growth, and survival, along with several housekeeping genes. Gene sets range between 5 and 70 genes depending on the assay (Table 2).

The instrument platforms also differ between assays. Oncotype DX, Breast Cancer Index, and EndoPrint use quantitative reverse transcription-polymerase chain reaction-based assays while MammaPrint, BluePrint, and TargetPrint use microarray-based assays. Differences in the genes these assays analyze and the methodologies they deploy could lead to varying outcomes for the same patient.

To illustrate, a study comparing the Prosigna and Oncotype DX assays in the same patient population found significant differences in risk classification (7). Despite the two assays showing concordance greater than 80% for the high-risk and low-risk RS groups, the study found substantial disagreement between the tests in the intermediate-risk RS group.

In the latter category, half of the patients were categorized as low-risk by the Prosigna test but high-risk by the Oncotype DX test. These differences could have translated to different treatment outcomes based on which test the oncologist ordered: Half of the patients in the intermediate-risk RS category might have received chemotherapy had they undergone Oncotype DX testing, whereas they would not have received chemotherapy based on the Prosigna test.

This discordance may confuse clinicians and affect patient outcomes. This makes it imperative for laboratorians and clinicians to critically evaluate the clinical validation data of these assays and to understand the differences between methods that might lead to discrepant risk classification.

Emerging MAAAs in Women’s Health

An emerging application of MAAA testing, especially in Europe, is to detect preeclampsia (PE). PE complicates 2%–3% of pregnancies and is a major cause of mortality and morbidity for mothers and babies. Severe PE can lead to preterm birth at <37 weeks’ gestation. The traditional approach to screening for PE is through maternal demographics and medical history, and this is the only approach recommended by ACOG. Risk factors include nulliparity, being older than age 40, having a body mass index (BMI) ≥35 kg/m2, conceiving via in-vitro fertilization, having a history of previous pregnancy with PE, family history of PE, chronic hypertension, chronic renal disease, diabetes mellitus, systemic lupus erythematosus, or thrombophilia (8).

In the United Kingdom, the National Institute for Health and Care Excellence (NICE) guidelines define high-risk of developing PE as having any one high-risk factor or any two moderate-risk factors (9). High-risk factors include a history of hypertensive disease in previous pregnancy, chronic kidney disease, autoimmune disease, diabetes mellitus, or chronic hypertension. Moderate-risk factors are first pregnancy, age older than 40 years, inter-pregnancy interval greater than 10 years, BMI at first visit of at least 35 kg/m2, or a family history of PE. The screening approaches from ACOG and NICE treat each risk factor as a separate screening test with additive detection rate and screen-positive rates. 

In contrast to traditional PE screening, MAAA testing has the potential to generate a patient-specific risk score. MAAA testing modifies a patient’s a priori risk with results of various biophysical and biochemical measurements to generate a PE screen risk using an approach similar to maternal serum screening for aneuploidy. Researchers have identified five useful biomarkers at 11-13 weeks’ gestation: mean arterial pressure (MAP), uterine artery pulsatility index (UtA-PI), serum placental growth factor (PlGF), and serum pregnancy associated plasma protein-A (PAPP-A).

One multicenter study conducted in the U.K. evaluated 20,168 singleton pregnancies at 11-13 weeks’ gestation (10). PE detection rates using MAAA-based screening performed better than current NICE recommendations. The PE detection rate by NICE recommendations was 40.8%, whereas MAAA using maternal factors, MAP, and PAPP-A was 54%; maternal factors, MAP, and PlGF 69%; and maternal factors, MAP, PlGF, and UtA-PI 82%.  

Another method developed by the Fetal Medicine Foundation uses a multivariable algorithm including maternal factors, MAP, UtA-PI, and PlGF. In a multicenter study of 35,948 singleton pregnancies at 11-13 weeks’ gestation, MAAA outperformed both the ACOG and NICE recommendations for PE detection (11). MAAA testing resulted in PE detection rates of 100% at <32 weeks, 75% at <37 weeks, and 43% at ≥37 weeks, with a 10% false positive rate. PE screening by ACOG recommendations had PE detection rates of 94% at <32 weeks, 90% at <37 weeks, and 89% at ≥37 weeks with a much higher false positive rate of 64%. Screening by NICE recommendations had lower detection rates of 41% at <32 weeks, 39% at <37 weeks, and 34% at ≥37 weeks with a 10% false positive rate.

Although the screening for PE using MAAA is not yet standard in clinical practice, preliminary data suggests that the addition of biomarkers in combination with maternal factors is superior to screening by maternal factors alone.

Conclusions

The field of MAAA testing is diverse and complex. Incorporating multiple biochemical or molecular analytes into algorithms with or without clinical information allows for a personalized risk assessment of a patient’s disease.

Given the potential clinical ramifications of MAAAs, FDA considers these tests to be high-impact and high-risk. Some are laboratory-developed and, depending on the company or laboratory offering the test, the performance of certain analytical or clinical parameters might not be available to the end user. Some companies and labs may not have data that reflect the potential risks of the test such as the rates of false negative and false positive results.

In contrast, FDA-approved MAAAs have documented clinical sensitivity and specificity. Nonetheless, differences between the tests may impact treatment decision-making, as noted earlier, and clinicians must be aware of these differences.

The increased attention to MAAAs suggests that more will enter practice in coming years. Yet reimbursement may remain a hurdle. While the Centers for Medicare and Medicaid Services does not pay for the algorithmic portion of MAAAs, the Protecting Access to Medicare Act that went into effect this year has increased payment for some of these tests. Currently there are at least 23 MAAAs with CPT codes assigned. Some of these consist of biochemical markers detected by immunoassay or mass spectrometry, with or without clinical information, while others use molecular genetic markers.

The scientific advancement of markers associated with complex illnesses and patients’ responsiveness to treatments, along with improved technologies in clinical and molecular pathology, will continue to drive development and implementation of MAAAs. It remains to be seen whether emerging MAAAs capable of refining diagnosis and guiding therapy will be routinely implemented in clinical laboratories and become the standard of care to improve women’s health.

Alicia Algeciras-Schimnich, PhD, DABCC, FADLM, is an associate professor of laboratory medicine and pathology, director of the clinical immunoassay laboratory, and chair of the clinical biochemistry and immunology division in the department of laboratory medicine and pathology at Mayo Clinic in Rochester, Minnesota.+Email: algeciras.alicia[at]mayo.edu

Katherine A. Turner, PhD, is a clinical chemistry fellow at Mayo Clinic in Rochester, Minnesota.+Email: turner.katherine1[at]mayo.edu

References

  1. Colon-Franco JM, Bossuyt PMM, Algeciras-Schimnich A, et al. Current and emerging multianalyte assays with algorithmic analyses-are laboratories ready for clinical adoption? Clin Chem 2018; doi: 10.1373/clinchem.2017.275677.
  2. Coleman RL, Herzog TJ, Chan DW, et al. Validation of a second-generation multivariate index assay for malignancy risk of adnexal masses. Am J Obstet Gynecol 2016;215:82.e1-11.
  3. Moore RG, McMeekin DS, Brown AK, et al. A novel multiple marker bioassay utilizing he4 and ca125 for the prediction of ovarian cancer in patients with a pelvic mass. Gynecol Oncol 2009;112:40-6.
  4. Grenache DG, Heichman KA, Werner TL, et al. Clinical performance of two multi-marker blood tests for predicting malignancy in women with an adnexal mass. Clin Chim Acta 2015;438:358-63.
  5. Practice bulletin no. 174: Evaluation and management of adnexal masses. Obstet Gynecol 2016;128:e210-26.
  6. Dabbs DJ. A comparison of breast cancer multianalyte assays with algorithmic analyses (maaa) for their net predictive/prognostic value. Clin Adv Hematol Oncol 2015;13:14-24.
  7. Kelly CM, Bernard PS, Krishnamurthy S, et al. Agreement in risk prediction between the 21-gene recurrence score assay (oncotype dx®) and the pam50 breast cancer intrinsic classifier™ in early-stage estrogen receptor-positive breast cancer. Oncologist 2012;17:492-8.
  8. ACOG. First-trimester risk assessment for early-onset preeclampsia. Committee opinion no. 638. Obstet Gynecol 2015;126:E25-7.
  9. National Collaborating Centre for Women’s and Children’s Health (UK). Hypertension in pregnancy: The management of hypertensive disorders during pregnancy. London: RCOG Press 2010.
  10. Tan MY, Wright D, Syngelaki A, et al. Comparison of diagnostic accuracy of early screening for pre-eclampsia by nice guidelines and a method combining maternal factors and biomarkers: Results of spree. Ultrasound Obstet Gynecol 2018.
  11. O’Gorman N, Wright D, Poon LC, et al. Multicenter screening for pre-eclampsia by maternal factors and biomarkers at 11-13 weeks’ gestation: Comparison with nice guidelines and ACOG recommendations. Ultrasound Obstet Gynecol 2017;49:756-60.
  12. Sparano JA, Gray RJ, Makower DF, et al. Adjuvent chemotherapy guided by a 21-gene expression assay in breast cancer. [Epub ahead of print] N Engl J Med. June 3, 2018 as 10.1056/NEJMoa1804710.