Morris KT, Pommier RF, Morris A, Schmidt WA, Beagle G, Alexander PW, Toth-Fejel S, Schmidt J, Vetto JT. Usefulness of the Triple Test Score for Palpable Breast Masses. Arch Surg. 2001;136(9):1008-1013. doi:10.1001/archsurg.136.9.1008
The triple test score (TTS) is useful and accurate for evaluating palpable breast masses.
Diagnostic test study.
University hospital multidisciplinary breast clinic.
Four hundred seventy-nine women with 484 palpable breast lesions evaluated by TTS from 1991 through July 2000.
Main Outcome Measures
Physical examination, mammography, and fine-needle aspiration were each assigned a score of 1, 2, or 3 for benign, suspicious, or malignant results; the TTS is the sum of these scores. The TTS has a minimum score of 3 (concordant benign) and a maximum score of 9 (concordant malignant). The TTS was correlated with subsequent histopathologic analysis or follow-up.
The TTS was prospectively calculated for each mass. Lesions with a TTS greater than or equal to 5 were excised for histologic confirmation, whereas lesions with scores less than or equal to 4 were either excised (n = 60) or followed clinically (n = 255).
All lesions with TTS less than or equal to 4 were benign on clinical follow-up, including 8 for which the fine-needle aspiration was the suspicious component. Of the 60 biopsied lesions, 51 were normal breast tissue, 4 showed fibrocystic change, 1 was a papilloma, and 4 were atypical hyperplasia. All lesions with a TTS greater than or equal to 6 (n = 130) were confirmed to be malignant on biopsy. Thus, a TTS less than or equal to 4 has a specificity of 100% and a TTS greater than or equal to 6 has a sensitivity of 100%. Of the 39 lesions (8%) with scores of 5, 19 (49%) were malignant, and 20 (51%) were benign.
The TTS reliably guides evaluation and treatment of palpable breast masses. Masses scoring 3 or 4 are always benign. Masses with scores greater than or equal to 6 are malignant and should be treated accordingly. Confirmatory biopsy is required only for the 8% of the masses that receive a TTS of 5.
THE TRIPLE test, initially described in 1975, is the evaluation of palpable breast masses by physical examination, mammography, and fine-needle aspiration (FNA) in women aged 40 years and older.1 The triple test is accurate and can replace open surgical biopsy for diagnosis when all 3 components are concordant; that is, all benign or all malignant. We confirmed this observation in 1995, resulting in a substantial reduction in the need for open surgical biopsies at our institution.2 However, we found that approximately 40% of cases in our experience had nonconcordant tests, thus requiring biopsy. In 1998, we improved the triple test with the concept of a triple test score (TTS), in which each component of the triple test was rated as benign, suspicious, or malignant, and assigned 1, 2, or 3 points, respectively. This scoring system further allowed us to reduce the number of required open biopsies when evaluating women aged 40 years or older with palpable breast masses.3
Our initial description of the TTS in 1998 has lead us to consider several issues regarding the reliability of the test. Specifically, we were interested in the safety of observing women with palpable masses or performing definitive therapy based only on this nonsurgical diagnostic test. In an effort to address these issues and further assess the accuracy of this diagnostic tool, we report our continued experience with the TTS in a multidisciplinary breast clinic.
Women aged 40 years or older referred to our multidisciplinary breast clinic between October 1, 1991, and February 1, 2000, for evaluation of palpable breast masses underwent prospective assessment by simultaneous physical examination, mammography, and FNA. All patients who underwent a complete TTS at our institution were entered into the study. In the early part of the series (October 1, 1991-June 30, 1997), patients who received a TTS of 4 or higher all underwent open biopsy. In the later part of the series (July 1997-February 2000), patients with scores of 4 were observed clinically, as were most patients with concordant benign tests, unless the patient or primary care provider requested a biopsy.
Each element of the TTS was given a score of 1, 2, or 3 points for benign, suspicious, or malignant findings. Physical examination results were determined by attending breast surgeons, staff radiologists reviewed mammography studies, and FNAs were performed and analyzed by experienced cytopathologists. Each element of the TTS was scored prospectively and independently. Patient data, the results of clinical follow-up, and pathology results were then collected for all patients and analyzed for this study.
Sensitivity and specificity rates for the TTS and its components were determined from the standard formulas, as follows: sensitivity = TP/(TP + FN), and specificity = TN/(TN + FP), where TP indicates true positive; TN, true negative; FP, false positive; and FN, false negative.4 The criterion standard was either the pathologic results or the results of clinical follow-up for the patients who did not receive biopsy.
There were 479 women with 484 breast masses evaluated. Mean patient age at time of evaluation was 49.5 years. The presenting complaint for 84% of the women was a new breast mass; 13% presented with an abnormal mammogram, and a corresponding mass was found in these women by the clinician on breast examination. Breast pain was the chief complaint for 1.8% of the women, with nipple discharge (0.2%) and other various complaints (1%) comprising the rest.
As summarized in Table 1, 315 breast masses (65%) had a TTS of 3 or 4 points. All were benign on biopsy or clinical follow-up. There were no cases of ductal carcinoma in situ (DCIS) diagnosed on either biopsy or follow-up in this group. The specifics regarding masses with a TTS of 4 are given in Table 2. Mean follow-up for patients with a TTS less than or equal to 4 was 14 months.
One hundred thirty breast masses (27%) had a TTS of 6 points or more. All were invasive carcinoma on biopsy. There were no cases of DCIS.
Thirty-nine breast masses (8%) had a TTS of 5 points. All had specimens obtained for biopsy, and 19 (49%) were malignant (including 1 case of DCIS). The specifics of these cases are given in Table 3. Counting "suspicious" scores as malignant and considering DCIS a malignancy, FNA (9 FP; 9 FN) and mammography (12 FP; 6 FN) findings were more accurate than findings from physical examination (17 FP; 7 FN). Thus, when the TTS was less than or equal to 4, it had a specificity of 100%, and when the TTS was greater than or equal to 6, it had a sensitivity of 100%. The sensitivity and specificity of the individual test elements for all the data are summarized in Table 4.
As of this report, we have used the TTS to evaluate 484 palpable breast masses in 479 women. After nearly doubling the size of our original series, the TTS retains its diagnostic sensitivity and specificity of 100% when the score does not equal 5 points (Table 4).3 This encompasses 92% of patients in our experience (Table 1). Therefore, we continue to regard the TTS as a powerful clinical tool that permits rapid, minimally invasive, and accurate diagnosis of palpable breast masses.
Since its inception, however, 3 major areas of clinical concern regarding the use of this scoring system have been identified by others and us. These have been as follows: (1) whether definitive therapy can be performed based on a TTS score of at least 6 points, without a confirmatory frozen section, especially when the score is equal to 6 with nonmalignant FNA findings; (2) whether it is safe to follow masses with a TTS of 4 (1 component suspicious) without biopsy; and finally (3) whether weighting the FNA would permit more definitive diagnosis for some patients with a TTS of 5 points, since FNA is the component with the highest sensitivity and specificity in previous reports and the current series (Table 4).2- 4
With regard to the first issue, the TTS has a diagnostic accuracy of 100% among patients with malignant breast masses. We do not believe that obtaining a confirmatory frozen section is necessary for these patients. The diagnostic accuracy of frozen section for breast masses ranges from 94.6% to 98.7%, with most incorrect diagnoses being FNs.5- 11 We do offer patients a frozen section biopsy intraoperatively, with the caveat that a FN frozen section diagnosis is actually more likely than a FP TTS; such an event might necessitate a second operation. Our department currently makes decisions regarding confirmatory biopsies or frozen sections based on individual consideration of the case by the attending surgeon and with an informed consent process from the patient. Confirmatory biopsies or frozen sections are not generally performed in patients whose TTS is greater than or equal to 6.
No woman in our series of 484 breast mass evaluations underwent a modified radical mastectomy or axillary dissection for DCIS or for a benign lesion. There was only 1 case of DCIS in this series, and that lesion received a TTS of 5, therefore leading to diagnostic biopsy prior to any treatment. In theory, it is possible that a woman could receive overtreatment of a benign or premalignant lesion with this scoring system, and we inform women of this possibility. However, in our 9 years of experience, no such errors have occurred. We are therefore confident of the safety of basing treatment decisions on this 3-part evaluation. In addition, it should be stressed that women in this study did not receive definitive treatment based solely on FNA. All therapeutic decisions were based on the results of the completed TTS. Finally, we believe our low rate of DCIS in this series was owing to the manner in which women were selected for TTS evaluation; that is, by presence of a palpable lump. As stated by Harris, "an abnormal mammographic result is the most common presentation of DCIS."12(p359) We therefore were less likely to include cases of DCIS in this study.
There were a total of 7 patients with scores of 6 in which all elements (physical examinations, mammography, and FNA) were suspicious but not definitely malignant. All 7 of these patients had invasive cancer. In addition, there were 2 cases in which the FNA was scored benign, but the patient received a TTS of 6 owing to the other test components being scored as malignant. These patients also had confirmed malignancy on lumpectomy, thus representing FN FNAs. These cases all illustrate the importance of scoring all the elements equally, as well as the reliability of a score of 6. Certainly, confirmatory biopsy with a core needle is an option for women with a TTS greater than or equal to 6; however, we do not advocate these in general, as they would only add a procedure and further expense to the workup of these patients. A TTS greater than or equal to 6 has correlated with 100% malignancy in this series of nearly 500 breast masses.
It is recognized that on FNA, an atypical fibroadenoma could potentially result in a score of 2 or 3, resulting in a TTS greater than or equal to 6 if 1 or more of the other components were labeled as suspicious.13 Obviously, this would lead to an incorrect diagnosis of malignancy and inappropriate treatment of a benign lesion. This has yet to occur in our sample with nearly 500 patients successfully evaluated. However, because of such a possibility, we inform all patients with a TTS greater than or equal to 6 that they are entitled to an open or core biopsy or frozen section prior to definitive treatment if they so choose.
A further advantage of our multidisciplinary breast clinic is that the cytopathologists are present to perform all FNAs, therefore reducing the frequency of nondiagnostic FNAs. They immediately assess the sample for adequacy and then repeat the FNA if the initial sample does not contain enough breast tissue to adequately place it in either the benign, suspicious, or malignant categories. Any woman with a mass for which an adequate FNA sample cannot be obtained is no longer a candidate for the TTS and undergoes a diagnostic biopsy. This situation, fortunately, is rare for our cytopathologists.
Regarding the second concern that patients whose masses receive a score of 4 (1 suspicious component) are only receiving clinical follow-up, we now report 39 cases with such scores with no malignancies found either on biopsy or clinical follow-up (Table 2). Four biopsies in this group did reveal atypical ductal hyperplasia. However, while we recognize that this lesion is a marker of increased risk for the later development of breast cancer, it is not a target lesion to be treated and not considered a malignancy in and of itself.14
Regarding the issue that the FNA should be weighted, we recognize that the FNA has the highest sensitivity and specificity of the 3 TTS components and provides a tissue diagnosis.2- 4 However, we have elected not to do this for several reasons. Five patients in our series received a benign reading of an FNA yet were found on biopsy to have an invasive cancer (ie, FN FNA findings). Of the 22 patients who received suspicious readings on FNA, 12 had benign lesions, and 10 had malignancies. It should be noted, however, that all patients in our study who received a malignant score for their FNA did, in fact, have a cancer on biopsy (n = 36). Thus, the FNA, while the most accurate individual test component (Table 4), does have its share of false results, which could weaken the reliability of the TTS if the FNA score were weighted.
Determining a way to diagnose patients with scores of 5 without biopsy continues to elude us. The lack of any major patterns in data from these patients and the frequent occurrences of both FPs and FNs for each test element (Table 3), along with small numbers, prevents us from making any recommendations within this subgroup. Thus, we believe masses with a TTS equal to 5 will continue to require biopsy until newer technology for evaluating them, such as improved breast magnetic resonance imaging techniques, are more established.15
The sensitivity and specificity of the FNA as performed in our department of pathology (Table 4) reveal this component of the TTS to be highly accurate, with only an 8% rate of FNs and 4% FPs in this patient population. Concern has been expressed that other pathology departments may not be able to reproduce this level of accuracy. This may or may not be the case, and we recommend each institution assure their own FNA accuracy rates are similar to ours before incorporating use of the TTS.
The TTS raises other intriguing clinical issues. Given the shift toward minimally invasive therapy for breast cancer, it has been noted by several authors that the sentinel lymph node (SLN) is more frequently identified and has higher accuracy when performed on patients whose tumors are still in place.16,17 By allowing the definitive diagnosis of malignancy prior to excision of the tumor, the TTS may enhance the ability to perform SLN biopsy in patients with palpable tumors. In addition, more patients may be able to have their entire surgical therapy performed in 1 operation, as opposed to having a biopsy, followed by a lumpectomy and an SLN biopsy or axillary node dissection.
Diagnosis by TTS may also offer an advantage to women who elect to have their cancer treated by a modified radical mastectomy. It is the standard of care for women who have had prior biopsy to have the biopsy site excised along with the mastectomy. If the lesion were far away from the nipple-areolar complex, this may require the excision of a considerable amount of skin. Women who have their diagnosis without undergoing open biopsy would be more often eligible for skin-sparing mastectomy with immediate reconstruction.
Finally, as we have reported in a previous article,2 the use of the TTS can result in charge reductions of at least $1412 per case. As that calculation was made at a time when 40% of our patients were still undergoing open biopsy for definitive diagnosis, use of the TTS can be expected to further lower patient charges.
In summary, we have used the TTS to successfully diagnose 484 palpable breast lumps and allow more than 200 women with benign lesions to avoid open diagnostic biopsy. Further, 130 women have had the option of proceeding with definitive therapy for their breast cancer without the need for separate biopsy.
Presented at the 72nd Annual Meeting of the Pacific Coast Surgical Association; Banff, Canada, February 18, 2001.
Corresponding author and reprints: John T. Vetto, MD, Section of Surgical Oncology, L223A, 3181 SW Sam Jackson Park Rd, Portland, OR 97201-3098 (e-mail: Vettoj@ohsu.edu).
James E. Goodnight, Jr, MD, PhD, Sacramento, Calif: It is an honor to discuss this work. I follow Dr Vetto's work very closely, had the honor to discuss it and go over it with him, and, clearly, he and his group have made this test work very well. We have 3 presentations this afternoon on how we efficiently get to a diagnosis and get on with the treatment of breast cancer.
The surgeon confronted with a patient who has a breast mass has several issues. First is the accuracy of diagnosis. The patient's welfare depends on this feature and, for that matter, so does everyone else's, except possibly the unlucky plaintiff attorney who is frozen out of a tight market. Second, efficiency of the diagnostic process. None of us have time to waste. Third, cost-effectiveness. Our current market demands this of us. Fourth, a definitive diagnostic result is essential. Both the surgeon and the patient need closure. Fifth, patient satisfaction, satisfaction of the referring MD, and sixth, getting paid for the work, all have to do with staying in business.
Drs Morris, Vetto, and their colleagues, by this presentation and their work of long standing, would seem to have satisfied themselves on the first 3 issues: accuracy, efficiency, and cost-effectiveness, and happily they are still in business. The triple test works well for them. They are a multidisciplinary breast clinic where the surgeon, mammographer, and cytopathologist are quite skillful and work side by side. Their results suggest that they communicate quite effectively. For this group, a nondefinitive result, that is, a score of 5, is distinctly uncommon, and a Hardway 6, that is, a score of 2, 2, and 2, is a definitive diagnosis of cancer. At issue is whether the obvious utility of the triple test is limited to a well-staffed multidisciplinary breast clinic, or is it applicable to other practice settings? The central question is how big will the group of 5 scores be, and will scores of 4 and 6 be definitive splits between benign and malignant in someone else's hands?
I will launch quickly into questions for the authors because I am very interested in their thoughts. (1) Is a dedicated cytopathologist who examines the patient with you essential for these crisp results? (2) Do your mammographers use standard mammographic reporting terminology such as the BIRADS system? (3) Does your pathologist and/or your mammographer know the surgeon's impression, that is, benign, suspicious, or malignant, when they evaluate the patient? (4) Do the surgeons, mammographers, and pathologists communicate verbally rather freely with each other on all or most of the patients?
I think presumptuously that the answer to each of these questions is yes, which I believe will indeed decrease the number of 5 scores and increase the sensitivity of the Hardway 6. But, obviously, I want to know your thoughts. I would also take advantage of the opportunity at the podium to ask a few practice-related questions. (1) You obviously do FNA [fine-needle aspiration] biopsies on patients where the surgeon and the mammographer have each graded the patient as 1, but I presume not all of these patients. How do you decide who gets the triple test or the FNA biopsy? (2) Even though it did not occur in your series, a patient with a score of 6 or greater may have a chance of having ductal carcinoma in situ. How do you decide before surgery whether you will do a sentinel lymph node biopsy of the axilla? (3) Do you ever use core needle biopsy? (4) Does your clinic return patients with a score of 3 and, in particular, 4 to their primary care provider for the on-going follow-up? And, finally, do your mammographers or do you routinely ultrasound very discrete smoothly rounded lesions that would seem to be breast cysts?
Howard Silberman, MD, Los Angeles, Calif: I am very much concerned on clinical and medicolegal grounds about the propriety of ignoring a suspicious FNA despite apparently nonworrisome physical examination and imaging. There is considerable subjectivity in the criteria for assigning the reading of "suspicious," depending upon the experience and self-confidence of the cytopathologist. The senior cytopathologist at the USC/Norris Cancer Hospital has reported to me that 80% to 90% of the lesions she regards as "suspicious" on FNA turn out to be malignant on excisional biopsy. I would like the authors to comment on this if they would. Similarly, to perform a mastectomy without a definitive histologic confirmation of a "suspicious" but not definitive FNA seems very risky business to me. We would always do a frozen section examination if not a prior core or excisional biopsy.
I. Benjamin Paz, MD, Duarte, Calif: This shows how important and how effective you can be by applying the triple test in the management of breast cancer. I am sure also that these patients were managed in a prospective breast conference where the concordance was established.
My question is, in ADH [atypical ductal hyperplasia], the authors said these patients do not require treatment. True, they don't require treatment for breast cancer. In the NSABP P1 trial, tamoxifen was shown to reduce 86% the incidence of breast cancer on these patients. Shouldn't these patients be offered that? If that is the case, should they be biopsied?
Richard Bold, MD, Sacramento: You specifically applied the triple test score to women over the age of 40, yet, it is women under the age of 40 who are the most common source of medicolegal complications from missed breast cancers. Why or cannot this triple test score be applied to women under 40 and, if it is a limitation in mammography, are there other imaging modalities that could supplant this for application of this scoring system to younger women with palpable lesions?
Dr Vetto: I want to thank the Association for the pleasure of presenting the data today, and I congratulate Dr Morris on an excellent presentation. I thank Dr Goodnight for his many fine comments and on his kindness in reviewing the manuscript for us, and I appreciate the other excellent questions from the other discussants.
As we expected, the questions on our work generally fall into 3 categories. The first are about the safety and accuracy of the test. The second group questions the transferability of the test to other clinical or practice settings. And finally, there were miscellaneous questions about other applications of the test. So let me take these in order.
First, the safety and accuracy questions. To begin with Dr Goodnight's, What about scores of greater than or equal to 6 equaling DCIS? We found no such cases in our series. Palpable DCIS is distinctly unusual, and I think we have shown that. As Dr Goodnight probably knows, Cox and others have described that using sentinel node for pure DCIS, one finds a higher-than 6% risk of positive lymph nodes, probably attributable to what is now called microinvasive DCIS. We think that, especially among patients who are choosing breast preservation therapy, the triple test lends itself very nicely to subsequent lumpectomy and sentinel lymph node biopsy regardless of whether one actually encounters the unusual case of palpable DCIS.
In terms of follow-up, we do send these patients back to the primary care providers, particularly if the score is concordant negative (3). In reports that I have published separately, we are working in Oregon to educate primary care providers on clinical breast examination so that patients come to the clinic with more clearly defined worrisome lesions and so that clinical breast exam can be dealt with more comfort in the community among patients who are either being primarily followed or have been to the clinic and now come back to their primary care provider. We have published previously the finding that teaching primary care providers clinical breast exam by means of a CME course such as is being sponsored by OHSU increases breast exam accuracy.
In terms of the role of ultrasound for obvious cysts, we do use in-clinic ultrasound for obvious cysts and aspirate them. Those patients were not in this series, and the use of ultrasound for triple testing is a separate issue, which I will discuss at the end.
I would like to thank Dr Silberman for emphasizing the very reason we developed the triple test score: because no 1 element of the triple test can be safely relied upon as the sole basis for diagnosis. This includes FNA, which, while the most accurate element, is still marred by false positives and false negatives. As Dr Silberman points out, by saying the FNA is "suspicious," the cytopathologist is saying that the lesion is either benign or malignant—he or she can't say which for sure. Thus, our scoring system takes this unreliable piece of information and places it in the context of the other 2 elements of the score to make it useful.
As you noticed, the suspicious FNA categories, ie, tests with scores of 4 (early in our experience) that had a suspicious FNA and all tests with scores of 5 that had suspicious FNA, did go to biopsy. In terms of scores of 6, the provider still has the option, and this is in the manuscript, to offer a frozen section prior to definitive therapy, but as we have shown you so far, we have found no malignancies in that setting. Our test clearly allows one to proceed with treatment planning and then do the frozen on the table if desired. In terms of frozen section for mastectomy, same comment. These are patients that I feel should probably get frozen section prior to proceeding on the table, especially those patients in whom one is planning bilateral mastectomy (where the patient has elected contralateral prophylactic mastectomy) based on a lower malignant score (6 and 7).
In terms of ADH, I believe that the data, when one looks at subset analysis of the NSABP's study, was fairly weak for the recommendation of tamoxifen. Yes, if we do encounter ADH, which is typically found by accident on a biopsy done for another reason, we do counsel that patient for tamoxifen. In my experience, many of those patients actually do not go on tamoxifen after they hear the data and/or meet with the genetic counselor.
In terms of transferability, Dr Goodnight asked about a dedicated cytopathologist. We showed in 1995 that obtaining the services of a dedicated cytopathologist is cost-effective, and I would recommend that to you. Do our radiologists use the BIRADS system? Yes, they do. In my opinion their task with the triple test score is the easiest. They simply read the x-ray as they normally would, and if the x-ray is given a BIRADS score of 1 or 2, then it receives a triple test score of 1. If they read a BIRADS score of 3 or 4, it receives a triple test score of 2, and if they read it as BIRADS 5, it receives a score of 3.
Are our evaluators blinded? Actually, they are. There is some communication that occurs during our multidisciplinary conference every week, but that is a minority of cases. The data that Dr Morris showed you was read mostly off of reports. So, yes, the test does work if the results are blinded. In fact, one of my favorite things to read in the report is "In the absence of clinical history, this radiographic (or this pathologic) impression is based on radiographic (or pathologic) grounds only." Independent scoring is the idea of the test.
In terms of the miscellaneous questions, who needs an FNA if both scores are benign? I would like to pause for a second and talk about that. If the patient comes to the clinic and is reported to have a palpable mass but no mass is found and the mammogram is normal, we document that "the patient has a negative doublet with no target for triple testing." If the patient has a lump that is present but it is felt to be benign and the imaging is negative, we offer the patient FNA, but we do not hard sell it if the patient is not interested. We simply document their decision. If the patient has either a suspicious examination or a suspicious imaging, we do recommend strongly to the patient that they proceed with the completion of the triplet and the triple test score because if the FNA is then at least suspicious, they will have a score of 5 and a subsequent 50% chance of malignancy.
Do we do core-needle biopsy? As originally described, the triple test includes either FNA or core biopsy. We like FNA because in our hands it actually is more accurate, allowing one to sample the entire lesion and not resulting in the sampling error that core can be associated with. It produces less bleeding. It can be read on site. It does not require a local anesthetic, and we in general favor cytopathology. In terms of transferability of the triple test score to the community, I think that this system is easily transferable to any multidisciplinary clinic in either a private or university setting. We would highly recommend it for those settings. If the setting is a multisurgeon private practice not associated with a large preexisting breast clinic, we would highly recommend obtaining the services of a skilled cytopathologist.
The clinical practice setting that I see most difficult for this is the solo practitioner, especially in a rural setting. I am of the opinion that those providers should consider referral of patients to multidisciplinary centers; certainly patient pressure leans in that direction. If the provider wishes to perform a triple test score, that is possible by performing one's own cytopathology and sending it off. That will, however, negate the benefit of same-day cytopathology. In fact nowadays a surgeon can even do their own radiographic interpretation, thanks to the American College of Surgeons Statement No. 29.
Finally, I would like to close with Dr Bold's question because it gets to our future plans. What about women younger than 40? We have previously published a modified triple test in which women of younger than screening age have a mass evaluated with ultrasound instead of with mammography. Dr Morris is in the process of culling the data on a modified triple test score for these women, and we hope to present that data in the future.