Using the Modified Magee Equation to Identify Patients Unlikely to Benefit From the 21-Gene Recurrence Score Assay (Oncotype DX Assay)

Materials and Methods

Breast cancer is the most frequent cancer in women and the second most frequent cause of cancer death among women within the United States.¹ More than half of breast cancer cases are estrogen receptor (ER) positive.² The combined antiestrogen therapy and chemotherapy have significantly reduced recurrence frequency and improved survival rate in certain populations of patients with ER-positive breast cancer.³ However, these therapies, especially chemotherapy, have toxic side effects; therefore, identifying patients who are more likely to benefit from chemotherapy is important. Clinicopathologic factors, including tumor size, histologic grade, lymph node status, and breast cancer biomarkers (ER, progesterone receptor [PR], and human epidermal growth factor receptor 2 [HER2]), have been used to determine if the patient needs systemic therapy and which therapies will be used.^4‐10

Recently, molecular tests have been clinically used to evaluate the patient’s recurrence risk and predict prognosis by assessing the genomic profile of the tumor.^11,12 Oncotype DX (Genomic Health, Redwood City, CA) is one such molecular test that has been widely accepted for clinical use within the United States and Canada. Oncotype DX, a reverse transcription–polymerase chain reaction assay of 21 genes, generates a recurrence score (RS) to predict the prognosis and chemotherapy benefits for ER-positive, HER2-negative patients with invasive breast cancer.^13,14 Oncotype DX quantifies the expression of 16 cancer-related genes after normalization using the remaining five genes to calculate the RS, which is reported as a number that is divided into low (<18), intermediate (18-30), or high (>30) recurrence risk categories.

Many studies have demonstrated that routine histopathologic and immunohistochemical variables can provide similar prognostic information to that of the Oncotype DX RS.^15‐27 This is not surprising because four of the 16 genes (ER, PR, HER2, and Ki-67) measured in Oncotype DX are assessed in the routine pathologic evaluation at the protein level by immunohistochemistry (IHC).²⁸ Tumor proliferation is also assessed in routine pathologic evaluation by counting mitosis as this is a component of tumor grade. Magee equations are derived by linear regression analysis using several pathologic variables and semiquantitative immunohistochemical results of ER, PR, HER2, and Ki-67 to calculate an RS that highly correlates with the Oncotype DX RS and provides similar information to that from Oncotype DX.^15,16

Results

Patients and Data Retrieval

Our goal was to independently examine the correlation between a modified Magee equation RS and Oncotype DX RS and investigate whether the modified Magee equation could categorize patients with ER-positive, HER2-negative breast cancer into the same low-risk or high-risk category as Oncotype DX.

Study Design

A total of 438 cases with available Oncotype DX RS were identified from the pathology files at The Ohio State University from 2010 to 2015. Demographic information, tumor size, Oncotype DX RS, ER, PR, and HER2 results were collected. All tumor H&E slides and immunohistochemical-stained slides were reviewed by breast subspecialized pathologists, with manual quantification of ER, PR, and HER2 (ER: clone SP1 [Spring, Pleasanton, CA], PR: clone PgR 636 [DAKO, Carpinteria, CA], HER2: 4B5 clone [Ventana, Tucson, AZ]). Fluorescence in situ hybridization (FISH) (US Food and Drug Administration–approved PathVysion HER2 DNA Probe Kit (Abbott Laboratories, Abbott Park, IL) consisting of a chromosome 17 centromeric probe and a separate HER2 probe) was performed on most cases with in-house core needle biopsy, and the FISH results were interpreted by fellowship-trained molecular pathologists. The Nottingham grade (NG) was calculated using the Nottingham modification of the Bloom-Richardson system.

Three Magee equations (1, 2, and 3) using different combinations of standard histopathologic variables (Nottingham Score [NS], ER, PR, HER2, Ki-67, and tumor size) have been previously reported to predict the Oncotype DX RS, available at the website of the pathology department at Magee-Womens Hospital of the University Pittsburgh Medical Center (http://path.upmc.edu/onlineTools/mageeequations.html). Since many cases in our study cohort did not have Ki-67 results, Magee equation 2 was used, which does not require a Ki-67 value to calculate the Magee RS. The H-score for ER or PR used in the original Magee equation 2 is calculated by adding the total of the stained percentage multiplied by intensity for all three different intensity grades (0-3) to obtain a value from 0 to 300. The H-score for ER and PR in our series was estimated to obtain a value from 0 to 300 by multiplying the predominant intensity (0-3) by the total stained percentage of cells (0-100). This modification of the Magee equation had been described previously.²⁹

There was an overall 67% (292/438) agreement between the modified Magee equation RS and the Oncotype DX RS Table 2. Following the criteria set by the Oncotype DX RS for low-, intermediate-, and high-risk categories, there were no two-step discordances (discordance between high and low risk) between any modified Magee equation RS and their paired Oncotype DX RS (Table 2). The Pearson correlation coefficient (r) for the modified Magee equation RS and Oncotype DX RS was 0.6645 (n = 438; P < .0001) with a mean coefficient of determination (r²) of 0.4416 Figure 1.

Figure 1

Open in new tabDownload slide

Correlation between the modified Magee equation recurrence score (RS) and Oncotype DX RS. The Pearson correlation coefficient (r) for the modified Magee equation RS and Oncotype DX RS was 0.6645 (n = 438; P < .0001). y = 0.3908x + 10.498. r²=0.4416.

Figure 1

Open in new tabDownload slide

Correlation between the modified Magee equation recurrence score (RS) and Oncotype DX RS. The Pearson correlation coefficient (r) for the modified Magee equation RS and Oncotype DX RS was 0.6645 (n = 438; P < .0001). y = 0.3908x + 10.498. r²=0.4416.

Table 2

Correlation Between the Modified Magee Equation RS and Oncotype DX RS

.	Oncotype DX RS .			Total No. .
Modified Magee Equation RS .	<18 .	18-30 .	>30 .
<18	186	62	0	248
18-30	55	95	29	179
>30	0	0	11	11
Total, No.	241	157	40	438

.	Oncotype DX RS .			Total No. .
Modified Magee Equation RS .	<18 .	18-30 .	>30 .
<18	186	62	0	248
18-30	55	95	29	179
>30	0	0	11	11
Total, No.	241	157	40	438

RS, recurrence score.

Open in new tab

Table 2

Correlation Between the Modified Magee Equation RS and Oncotype DX RS

.	Oncotype DX RS .			Total No. .
Modified Magee Equation RS .	<18 .	18-30 .	>30 .
<18	186	62	0	248
18-30	55	95	29	179
>30	0	0	11	11
Total, No.	241	157	40	438

.	Oncotype DX RS .			Total No. .
Modified Magee Equation RS .	<18 .	18-30 .	>30 .
<18	186	62	0	248
18-30	55	95	29	179
>30	0	0	11	11
Total, No.	241	157	40	438

RS, recurrence score.

Open in new tab

Statistical Analysis

Discussion

Correlation and Concordance Between Modified Magee Equation RS and Oncotype DX RS

All cases were stratified into six groups based on different combinations of the modified Magee equation RS and Oncotype DX RS: (1) modified Magee equation RS less than 18/Oncotype DX RS less than 18, (2) modified Magee equation RS less than 18/Oncotype DX RS of 18 to 30, (3) modified Magee equation RS of 18 to 30/Oncotype DX less than 18, (4) modified Magee equation RS of 18 to 30/Oncotype DX of 18 to 30, (5) modified Magee equation RS of 18 to 30/Oncotype DX more than 30, and (6) modified Magee equation RS more than 30/Oncotype DX more than 30 Table 3.

One hundred percent of cases with the modified Magee equation RS less than 18 (n = 248) had an Oncotype DX RS less than 30 (range, 0-29), including 186 cases with an Oncotype DX RS less than 18 (group 1) and 62 cases with an Oncotype DX RS between 18 and 30 (group 2) (Table 3). One hundred percent of cases with the modified Magee equation RS greater than 30 (n = 11) had an Oncotype DX RS more than 30 (range, 33-61; group 6) (Table 3).

There were 186 (42.5%) cases in this group. Approximately 30% of cases were low-grade histologic types (24.2% of lobular, 4.3% of mucinous, and 1.1% of tubular carcinoma). Three (2%) cases were PR negative, and none were ER negative by Oncotype DX. None of the cases were ER or PR negative by IHC. Three (1.5%) cases showed discordant PR results between Oncotype DX and IHC. Forty-eight (26%) cases had lymph node(s) positive for metastatic carcinoma (either micrometastasis or macrometastasis).

There were 62 (14%) cases in this group. Approximately 26% of cases were low-grade histologic types (23% of lobular, 2% of mucinous, and 1.6% of tubular carcinoma). Four (6%) cases were PR negative, and none were ER negative by Oncotype DX. None of the cases were ER negative, but one (2%) case was PR negative by IHC. Five (8.1%) cases showed discordant PR results between Oncotype DX and IHC. Twelve (19%) cases had lymph node(s) positive for metastatic carcinoma (either micrometastasis or macrometastasis). Between group 1 and group 2, the following parameters were found to be significantly different: modified Magee equation RS, Oncotype DX RS, ER/PR values by Oncotype DX, HER2 value by Oncotype DX, and ER/PR scores by IHC (P < .05). The percentage of cases with discordant PR results between Oncotype DX and IHC showed a marginal difference between these two groups (1.5% vs 8.1%) (Table 3).

There were 55 (12.6%) cases in this group. Approximately 29% of cases were low-grade histologic types (27% of lobular and 2% of mucinous carcinoma). Four (7%) cases were PR negative, and none were ER negative by Oncotype DX. None of the cases were ER negative, but four (7%) cases were PR negative by IHC. Eight (15%) cases showed discordant PR results between Oncotype DX and IHC. Fifteen (27%) cases had lymph node(s) positive for metastatic carcinoma (either micrometastasis or macrometastasis). Between group 3 and group 2, the following parameters were found to be significantly different: modified Magee equation RS, Oncotype DX RS, ER values by Oncotype DX, HER2 value by Oncotype DX, NG and all three grading components (tubule, nuclear pleomorphism, and mitosis), and PR score by IHC (P < .05) (Table 3).

There were 95 (21.7%) cases in this group. Approximately 17% of cases were low-grade histologic types (16% of lobular and 1% of mucinous carcinoma). Twenty-five (26%) cases were PR negative, and none were ER negative by Oncotype DX. None of the cases were ER negative, but 15 (16%) cases were PR negative by IHC. Nineteen (20%) cases showed discordant PR results between Oncotype DX and IHC. Twenty-three (24%) cases had lymph node(s) positive for metastatic carcinoma (either micrometastasis or macrometastasis). Between group 4 and group 3, the following parameters were found to be significantly different: modified Magee equation RS, Oncotype DX RS, ER/PR values by Oncotype DX, PR-positive percentage by Oncotype DX, and HER2 value by Oncotype DX (P < .05) (Table 3).

There were 29 (6.7%) cases in this group. Three (10%) cases were lobular carcinoma (two classic and one pleomorphic lobular carcinomas), and all others were ductal carcinoma. One (3%) case was ER negative and 19 cases (66%) were PR negative by Oncotype DX. None of the cases were ER negative, but six (21%) cases were PR negative by IHC. Thirteen (45%) cases showed discordant PR results between Oncotype DX and IHC. Seven (24%) cases had lymph node(s) positive for metastatic carcinoma (either micrometastasis or macrometastasis). Between group 5 and group 4, the following parameters were found to be significantly different: modified Magee equation RS, Oncotype DX RS, PR-positive percentage by Oncotype DX, PR values by Oncotype DX, percentage of cases with discordant PR results, and HER2 value by Oncotype DX (P < .05) (Table 3).

Clinicopathologic Features of Six Groups With Different Combinations of Modified Magee Equation RS and Oncotype DX RS

There were 11 (2.5%) cases in this group. The mean age for this group was 47.9 years (range, 31-61 years), significantly lower than the age in other groups. All cases were ductal carcinomas. Seven (64%) cases were ER negative and nine (82%) were PR negative by Oncotype DX. None of the cases were ER negative, but five (45%) cases were PR negative by IHC. Five (45%) cases showed discordant PR results between Oncotype DX and IHC. Three (27%) cases had lymph node(s) positive for metastatic carcinoma (either micrometastasis or macrometastasis). Between group 6 and group 5, the following parameters were found to be significantly different: age, modified Magee equation RS, Oncotype DX RS, ER values by Oncotype DX, ER-positive percentage by Oncotype DX, and NG and two grading components (nuclear pleomorphism and mitosis) (P < .05) (Table 3).

Cases With Modified Magee RS Less Than 11 or Modified Magee RS More Than 30 and Cost Analysis

One hundred percent of cases with the modified Magee equation RS less than 11 (n = 40) had an Oncotype DX RS less than 18, and 100% of cases with the modified Magee equation RS more than 30 (n = 11) had an Oncotype DX RS more than 30 Table 4. These cases accounted for 11.6% of all cases with Oncotype DX test and had been correctly predicted by the modified Magee equation to have the corresponding Oncotype DX RS category. The estimated cost saving from these patients by not performing Oncotype DX would be $221,850, based on the current list price of $4,350 for Oncotype DX.

The American Society of Clinical Oncology clinical practice guideline recommends the clinician may use the 21-gene RS (Oncotype DX RS) to guide decisions on adjuvant systemic chemotherapy if a patient has ER-positive, HER2-negative breast cancer, and the Oncotype DX assay has been widely accepted in clinical practice across the United State and Canada. Since four of the 16 genes (ER, PR, HER2, and Ki-67) measured as part of the Oncotype DX panel are assessed at the protein level by IHC in the routine diagnostic evaluation of breast cancer, studies have suggested that standard histopathologic variables (including tumor grade and mitotic activity), together with biomarkers evaluated by IHC, can provide information similar to that provided by the Oncotype DX RS.^15‐27 The Magee equation is one of such approaches to predict Oncotype DX RS by using different combinations of NG, Ki-67, or tumor size, in addition to semiquantitative ER, PR, and HER2.^15,16 In this study, we examined the correlation between the modified Magee equation RS and Oncotype DX RS and investigated whether the modified Magee equation could categorize patients with ER-positive breast cancer into either low-risk or high-risk groups concordant with the Oncotype DX assay.

As with previous studies,^16,29 our current investigation demonstrated excellent concordance between the modified Magee equation RS and Oncotype DX RS, with an overall 67% (292/438) agreement and no “two-step” discordance (discordant high and low RS). The concordance was even higher (76%, 197/259) excluding the intermediate modified Magee equation RS. Cases with higher Oncotype DX RS typically had a higher modified Magee equation RS, tumor grade, and mitotic activity and a lower ER and PR. Cases with lower Oncotype DX RS typically had a lower modified Magee equation RS, tumor grade, and mitotic activity and a higher ER and PR.

References

The modified Magee equation, which uses standard histopathologic and immunohistochemical variables, could identify a portion of patients with ER-positive breast cancer who may not need Oncotype DX for determining an RS. Indeed, our data demonstrated that all cases with a modified Magee equation RS greater than 30 had an Oncotype DX RS of more than 30, and 100% (n = 40) of cases with a modified Magee equation RS of 11 or less had an Oncotype DX RS less than 18. These cases accounted for 11.6% of all cases with the Oncotype DX test and had been correctly predicted by the modified Magee equation to have a corresponding Oncotype DX RS category. Therefore, patients with a modified Magee equation RS less than 11 or more than 30 may not benefit from an Oncotype DX test. The estimated cost saving from these patients would be $221,850, based on the current list price of $4,350 for the Oncotype DX. This is only for a single institution with a moderate volume of breast cancer cases over a 6-year period. Similar cost saving analysis was also calculated in a previous study using a screening algorithm with the combination of modified Magee equations, histologic criteria, and biomarker results to identify potential cases unlikely to benefit from the Oncotype DX.²⁹ Our data support this idea with a simplified process and still were able to identify up to 12% of Oncotype DX eligible cases that may not need Oncotype DX.

As with other studies,^14,16,25,30 our data also demonstrated that most tumors (92.5%, 37/40) with a high Oncotype DX RS were invasive ductal carcinomas (IDCs). Furthermore, all tumors with both a high modified Magee equation RS and an Oncotype DX RS were IDCs. However, the number of tumors with a high Oncotype DX was small (40 cases, 9.1% of total cases), which was probably caused by selection bias with underrepresented apparent high-risk or low-risk patients. Indeed, the previous study by Turner et al²⁹ also demonstrated a similar low percentage of tumors with a high Oncotype DX RS (11.4%, 29/255). Therefore, further studies with more tumors with a high Oncotype DX RS are warranted. All invasive lobular carcinoma (ILC) cases had a modified Magee equation RS less than 30 (100%, 92/92), and all ILC cases except three cases had an Oncotype RS less than 30 (96.7%, 89/92). These three ILC cases had an Oncotype DX RS of 32, 33 and 36; had a modified Magee equation RS of 20.3, 26.4, and 20.0; and were negative PR by Oncotype DX. Two also had negative PR by IHC, and one case had a very low PR H-score of 5. Two were classical type and one of them was pleomorphic type. The overall mean modified Magee equation RS for all ILCs was 16.7, and the mean Oncotype DX RS was 15.9. The findings are consistent with those from other studies showing only rare cases of ILCs with a high Oncotype DX RS,^30‐33 suggesting classical ILCs may not benefit from Oncotype DX testing. Similarly, neither of the two other specific types of breast carcinomas (three invasive tubular carcinomas and 11 invasive mucinous carcinomas) had a high Oncotype DX RS or a high modified Magee equation RS. Among these 14 cases, 10 cases had both a low modified Magee equation RS and a low Oncotype DX RS, two cases had a low modified Magee equation RS and an intermediate Oncotype DX RS, one case had an intermediate modified Magee equation RS and low Oncotype DX RS, and one case had both RSs as intermediate. Both tubular carcinoma and mucinous carcinoma are considered indolent tumors having a good prognosis. Although our findings may implicate that performing Oncotype DX RS on these special types of tumors is less likely to be useful to guide clinical management, the number of these special types (tubular and mucinous types) in our study was too low to draw a definitive conclusion, and further studies with large cohorts of these specific types are warranted.

Our study cohort was limited by bias toward patients with an indeterminate clinical risk assessment because the selection of cases for Oncotype DX was based on the oncologists’ clinical judgment. Therefore, apparent low-risk or high-risk patients are likely to be underrepresented. Another limitation of our study is the lack of recurrence or overall survival data because the study cohort only included relatively recent cases that did not have sufficient time for clinical follow-up. In our study, ER and PR IHCs were evaluated by a semiquantitative manner. Although such a method might be subjective with a variable range, the overall quality and interobserver agreement for interpretation of these results were ensured by a quality assurance program and a mandatory proficiency test implemented in our institution. In addition, low-grade carcinoma cases with weak ER staining/negative PR staining and high-grade cases with positive ER/PR were reviewed in the current study.

In summary, our data provided more evidence to support that the modified Magee equation RS correctly predicts Oncotype DX RS for most cases, and there should be little difference between these two RS risk categories. However, if a major difference occurs, the pathology, biomarkers, and the block sent for Oncotype DX testing should be thoroughly investigated. Furthermore, the modified Magee equation derived from standard histologic variables and breast biomarker results can be used to stratify patients who are eligible for Oncotype DX testing, identifying up to 12% of patients who are unlikely to benefit from this test and saving unnecessary expenses.