Symptoms include hoarseness, dysphagia, discomfort, pain, pressure, and difficulty breathing, as well as symptoms of hypothyroidism and hyperthyroidism, with (+) indicating present and (−) indicating absent. AUS indicates atypia of undetermined significance; FN, suspicious for follicular neoplasm; FNA, fine-needle aspiration; MNG, multinodular goiter; and SHCN, suspicious for Hürthle cell neoplasm.
aIf preoperative or intraoperative suspicious lymph nodes are identified, a total thyroidectomy with therapeutic central compartment neck dissection would be performed.
The results are shown in 140 patients who underwent molecular testing.
Salem I. Noureldine, Alireza Najafian, Patricia Aragon Han, Matthew T. Olson, Dane J. Genther, Eric B. Schneider, Jason D. Prescott, Nishant Agrawal, Aarti Mathur, Martha A. Zeiger, Ralph P. Tufano. Evaluation of the Effect of Diagnostic Molecular Testing on the Surgical Decision-Making Process for Patients With Thyroid Nodules. JAMA Otolaryngol Head Neck Surg. 2016;142(7):676–682. doi:10.1001/jamaoto.2016.0850
Diagnostic molecular testing is used in the workup of thyroid nodules. While these tests appear to be promising in more definitively assigning a risk of malignancy, their effect on surgical decision making has yet to be demonstrated.
To investigate the effect of diagnostic molecular profiling of thyroid nodules on the surgical decision-making process.
Design, Setting, and Participants
A surgical management algorithm was developed and published after peer review that incorporated individual Bethesda System for Reporting Thyroid Cytopathology classifications with clinical, laboratory, and radiological results. This algorithm was created to formalize the decision-making process selected herein in managing patients with thyroid nodules. Between April 1, 2014, and March 31, 2015, a prospective study of patients who had undergone diagnostic molecular testing of a thyroid nodule before being seen for surgical consultation was performed. The recommended management undertaken by the surgeon was then prospectively compared with the corresponding one in the algorithm. Patients with thyroid nodules who did not undergo molecular testing and were seen for surgical consultation during the same period served as a control group.
Main Outcomes and Measures
All pertinent treatment options were presented to each patient, and any deviation from the algorithm was recorded prospectively. To evaluate the appropriateness of any change (deviation) in management, the surgical histopathology diagnosis was correlated with the surgery performed.
The study cohort comprised 140 patients who underwent molecular testing. Their mean (SD) age was 50.3 (14.6) years, and 75.0% (105 of 140) were female. Over a 1-year period, 20.3% (140 of 688) had undergone diagnostic molecular testing before surgical consultation, and 79.7% (548 of 688) had not undergone molecular testing. The surgical management deviated from the treatment algorithm in 12.9% (18 of 140) with molecular testing and in 10.2% (56 of 548) without molecular testing (P = .37). In the group with molecular testing, the surgical management plan of only 7.9% (11 of 140) was altered as a result of the molecular test. All but 1 of those patients were found to be overtreated relative to the surgical histopathology analysis.
Conclusions and Relevance
Molecular testing did not significantly affect the surgical decision-making process in this study. Among patients whose treatment was altered based on these markers, there was evidence of overtreatment.
The clinical importance of thyroid nodules primarily rests with the need to exclude ones that may harbor malignancy. Therefore, the challenge for any diagnostic test is to sustain a high sensitivity such that cancers are not missed, while achieving a high negative predictive value to avoid surgery in patients whose nodules are benign. This achievement is usually accomplished by obtaining a detailed history and physical examination, in addition to ultrasound of the neck. The information attained will direct subsequent evaluation, including the need for fine-needle aspiration (FNA) biopsy assessment.1 Standardized interpretation of FNA cytology has improved substantially since the implementation of the tiered Bethesda System for Reporting Thyroid Cytopathology (BSRTC) classification scheme.2,3 Yet, approximately one-third of all thyroid aspirates yield indeterminate or suspicious cytology according to the BSRTC classification, with an estimated malignancy rate ranging from 5% to 75%.4,5
Owing to the lack of a definitive diagnosis for cytologically indeterminate or suspicious thyroid nodules, diagnostic surgery is required in most cases to establish a histopathological diagnosis and reconcile the behavior of these nodules. However, only 16% to 26% of surgically resected cytologically indeterminate nodules and 75% of cytologically suspicious nodules prove to be malignant.6 Such diagnostic operations, with their associated expense, risk, and patient burden, could be avoided if FNA procedures could reliably establish the preoperative diagnosis of a benign nodule. Conversely, both unnecessary and multiple surgical procedures could potentially be avoided with a more accurate preoperative diagnosis of a malignant thyroid nodule.
Much progress has been made in understanding the molecular mechanisms of thyroid cancer over the past 2 decades.7 This knowledge has led to the identification and clinical use of novel diagnostic molecular markers as preoperative adjuncts to improve risk stratification and hence tailor more effective management strategies in patients with thyroid nodules.8,9 While these diagnostic molecular tests appear promising in differentiating benign from malignant thyroid nodules, their true effect on the preoperative surgical decision-making process has yet to be fully demonstrated. Preliminary results indicate that molecular testing is overused, is misinterpreted, and does not alter the surgical management in most patients in whom this testing has been done.10,11 This occurrence is likely due to the lack of specific guidelines from professional societies on the proper use of these markers12 and, most important, the fact that current management strategies of nodular thyroid disease usually take into account a myriad of clinical considerations beyond cytopathological findings alone.1 Assessment of diagnostic molecular marker test results in the context of these clinical considerations is necessary to evaluate their true effect on the surgical decision-making process.
Question Does molecular profiling of thyroid nodules affect the surgical decision-making process?
Findings In this case-control study, the surgical management plan was altered in only 7.9% (11 of 140) of patients with thyroid nodules owing to the diagnostic molecular test result. Most of those patients were found to be overtreated.
Meaning Diagnostic molecular testing should not be used reflexively or in isolation to drive clinical decision-making in patients with thyroid nodules. These tests should only be considered once patient, clinical, radiological, and cytological factors are weighed in a balanced manner and it is determined that a molecular test result would appropriately alter the treatment plan.
Under The Johns Hopkins University School of Medicine institutional review board approval and waiver of informed consent, we conducted a prospective study to evaluate all consecutive patients who were seen by 5 thyroid surgeons (J.D.P., N.A., A.M., M.A.Z., and R.P.T.) for surgical consultation at The Johns Hopkins Hospital between April 1, 2014, and March 31, 2015. Patients with thyroid nodules who underwent diagnostic molecular testing before being referred to 1 of our 5 thyroid surgeons were included in this study. In all cases, molecular profiling of the FNA specimens obtained from these patients was requested by either the outside referring physician or an endocrinologist or internist at The Johns Hopkins University before surgical consultation. Patients who had a history of thyroid surgery or who had undergone molecular profiling of a thyroid nodule in the past were excluded. Patients with thyroid nodules who did not undergo molecular testing and were seen for surgical consultation during the same period served as a control group.
As a standard of care, all outside thyroid FNA specimens were routinely read by one of the cytopathologists at The Johns Hopkins University. The final in-house cytological diagnoses were categorized according to the BSRTC classification, as described previously.13 Categorical diagnoses I through VI included nondiagnostic (I), benign (II), atypia of undetermined significance (III), suspicious for follicular neoplasm or suspicious for Hürthle cell neoplasm (IV), suspicious for malignancy (V), and malignant (VI). Diagnostic molecular marker tests included (1) an RNA-based gene expression classifier (GEC) (Afirma; Veracyte Inc), (2) a DNA-based somatic mutation panel (SMP) (Asuragen [Asuragen Clinical Laboratory], ThyroSeq [CBLPath], or Quest [Quest Diagnostics]), and (3) BRAF (OMIM 164757) V600E mutation analysis with or without RAS (OMIM 164790 for NRAS, OMIM 190070 for KRAS, and OMIM 190020 for HRAS) or RET/PTC (OMIM 164761) alterations.8,9,14- 18
Before initiating the study, a surgical management algorithm was developed and published after peer review.10 It incorporated individual BSRTC classifications (in-house cytological diagnosis), in addition to clinical, laboratory, and radiological findings. The algorithm was created by consensus from the dedicated thyroid surgeons and authors (S.I.N., A.N., M.T.O., J.D.P., N.A., A.M., M.A.Z., and R.P.T.) to formalize their decision-making process in managing patients with thyroid nodules (Figure 1). Creating a management algorithm allowed us to determine whether or not a molecular test result had an effect on the decision-making process in real time during the surgical consultation. The actual management undertaken by the surgeon was then compared with the recommended management option within the corresponding arm in the algorithm. If the surgeon’s management recommendation was the same as expected from the algorithm, the molecular test result was considered to have no effect on the surgical decision-making process. Conversely, if the surgeon’s management recommendation was different from what was dictated by the algorithm, the specific reasons for the change in management and extent of treatment were prospectively recorded in real time during the consultation.
A mobile device data entry user interface was developed using a software program (FileMaker Pro 13; FileMaker, Inc) and directly linked to a secure data storage server to allow data collection and analysis in real time during the surgical consultation. Using the standardized management algorithm, a calculator was developed with a unique patient identifier for each entry on a tablet (iPad; Apple Inc), which allowed for data entry during the consultation. If the surgeon’s management recommendation matched the one suggested by the algorithm, the case was signed and stored, indicating no change in the decision-making process. If the surgeon’s management recommendation was different from what the algorithm depicted, the software prompted entry of the specific reasons for the change in management.
The specific reasons for the change in management and extent of treatment were recorded. Demographic, clinical, laboratory, and radiological data were also recorded. For patients who proceeded to surgery, the final histopathology report was reviewed for each case and compared with the preoperative in-house cytology report for direct correlation between samples, as well as to identify the histopathological nature of the biopsied nodule. Incidental malignant thyroid neoplasms were not included in the analysis.
To evaluate the appropriateness of any change in management, the surgical histopathology diagnosis was correlated with the surgery performed. If the nodule was histologically malignant and the change in management was from a less extensive to a more extensive surgery (eg, from a hemithyroidectomy to a total thyroidectomy), the change was deemed appropriate. Conversely, if the nodule was histologically benign and the change in management was from a more extensive to a less extensive surgery, the change was also deemed appropriate. Any other change was deemed inappropriate.
Data analysis was performed using statistical software (Stata SE, version 13.1; StataCorp LP). Logistic regression models were constructed to analyze bivariable associations of the occurrence of molecular testing with demographics, clinical characteristics, and changes in outcome. Next, a multivariable logistic regression model was constructed to determine independent predictors of the occurrence of molecular testing and changes in outcome. For the multivariable model, variables found to have a significant association with the occurrence of molecular testing on bivariable analysis were selected for inclusion. The value reported for each variable is an odds ratio (OR) with a 95% CI. All reported P values are 2-sided, and a level of .05 was used to evaluate statistical significance. All values are expressed as means (SDs), ranges, or absolute numbers.
Over the 1-year period, 703 patients with nodular thyroid disease and no history of thyroid surgery were referred for surgical consultation. Fifteen of those patients had missing data (no in-house FNA cytology review or no FNA biopsy on presentation) and were excluded. The remaining 688 patients were analyzed. Fifty-five patients (8.0%) were referred by an endocrinologist or internist at The Johns Hopkins University and the remaining 633 patients (92.0%) were referred by an outside physician.
The mean (SD) age of 688 patients (559 female and 129 male) at presentation was 51.1 (14.8) years (age range, 9-93 years). Overall, 20.3% (140 of 688) underwent molecular testing before referral, and 79.7% (548 of 688) did not. Table 1 summarizes the characteristics of these 140 patients. One patient underwent molecular testing for 2 different nodules, and 5 patients underwent 2 different molecular tests for the same nodule. Overall, molecular profiling of the thyroid nodules included 128 GEC tests, 9 SMP tests, and 9 point mutation (ie, BRAF and RAS) tests. When these diagnostic molecular studies were performed, 1.4% (2 of 140) of the nodules were cytologically nondiagnostic, 7.1% (10 of 140) were benign, 66.5% (93 of 140) were atypia of undetermined significance, 14.3% (20 of 140) were suspicious for follicular neoplasm, 6.4% (9 of 140) were suspicious for Hürthle cell neoplasm, 2.1% (3 of 140) were suspicious for malignancy, and 2.1% (3 of 140) were malignant.
On bivariable analysis, patients with referral from the community (OR, 0.39; 95% CI, 0.22-0.70; P = .002), with larger nodules (OR, 0.83 per each centimeter increase in size; 95% CI, 0.73-0.95; P = .007), with compressive symptoms (OR, 0.42; 95% CI, 0.26-0.66; P < .001), and with hyperthyroidism (OR, 0.21; 95% CI, 0.06-0.74; P = .02) were less likely to undergo molecular testing than their counterparts. Men were 61% more likely to undergo molecular testing compared with women (OR, 1.61; 95% CI, 1.03-2.51; P = .04). On multivariable analysis, the likelihood of undergoing molecular testing was only independently based on referral from the community (OR, 0.42; 95% CI, 0.23-0.77; P = .005), the presence of compressive symptoms (OR, 0.55; 95% CI, 0.33-0.92; P = .02), and hyperthyroidism diagnosis (OR, 0.25; 95% CI, 0.07-0.90; P = .03).
The management dictated by the surgeon deviated from the surgical treatment algorithm in only 18 of 140 (12.9%) with molecular testing and in 56 of 548 (10.2%) without it (P = .37). In the 56 control patients, the surgical management plan deviated from the algorithm as a result of the following factors: the ultrasound characteristics (n = 13), the presence of cytologically benign contralateral nodules (n = 2), age and comorbidities (n = 9), patient preference (n = 38), nodule size (n = 16), rapid nodule growth (n = 2), family history of thyroid cancer (n = 1), outside FNA cytology diagnosis (n = 1), and in-house FNA cytology (n = 5). In the group with molecular testing, the management was altered as a result of the ultrasound characteristics (n = 4), the presence of cytologically benign contralateral nodules (n = 5), patient preference (n = 11), nodule size (n = 3), and molecular testing results (n = 11).
On simple logistic regression, uninodular status and nodule size were associated with an increased odds of a change in the treatment plan. Among those with molecular testing, uninodular status (OR, 26.30; 95% CI, 2.70-256.00; P = .005) and nodule size (OR, 1.66 per each centimeter increase; 95% CI, 1.11-2.48; P = .02) were associated with an increased odds of a change in the treatment plan after controlling for age, sex, uninodularity, and nodule size.
Of the 11 of 140 patients (7.9%) whose treatment was altered as a result of the molecular test result, follow-up was recommended for 3 patients, and a total thyroidectomy in lieu of a hemithyroidectomy was suggested for the other 8. Relative to the histopathology analysis, the management was appropriately improved in only 1 of 8 patients (12.5%) with an indeterminate nodule as a result of the molecular test result (GEC suspicious). This patient’s nodule was found to harbor a follicular variant of papillary thyroid cancer (pT1bNxMx) with a background of chronic lymphocytic thyroiditis. Conversely, the surgical management was altered inappropriately in the remaining 7 of 8 patients (87.5%) who had indeterminate cytology and a suspicious GEC result. On histopathology, the nodules were found to be follicular adenoma (n = 5), adenomatoid nodule (n = 1), and Hürthle cell adenoma (n = 1), with no incidental thyroid cancers identified in the entire specimen. The change in management was due to concern about the higher probability of malignancy with a suspicious molecular result or due to the influence of the referring endocrinologist and his or her interpretation of a suspicious GEC result. Table 2 summarizes the cytology diagnoses, molecular testing results, and final histopathology diagnoses for the 140 patients who underwent molecular testing.
After evaluating the cases in which the molecular test result did not change the surgical management, we found that the GEC test did not add any benefit to the decision-making process in the following settings: (1) the FNA biopsy yielded material that was nondiagnostic, benign, suspicious for malignancy, or malignant on cytological evaluation; (2) nodules were present with strong clinical or radiological findings concerning for a malignant thyroid neoplasm; (3) indeterminate nodules measured greater than 4 cm and were causing compressive symptoms for which surgery would be recommended regardless; and (4) patients preferred surgery and were uncomfortable with conservative management of a nodule irrespective of the GEC result. The SMP molecular marker testing did not add benefit to the surgical decision-making process when (1) the nodule was malignant or benign on cytology evaluation, (2) regional lymph node metastases were present, (3) patients had clinical considerations for a total thyroidectomy (eg, multinodular goiter, hypothyroidism, strong family history of thyroid cancer, and previous radiation exposure), and (4) patients had a strong preference for a total thyroidectomy. In these cases, the molecular test was not clinically indicated (Figure 2).
In a recently published study,11 our group reviewed our surgical experience in managing patients who underwent GEC testing of a thyroid nodule before being referred for surgical consultation. We found that in most patients the GEC test was not used for its intended purpose, namely, as a negative predictor of malignancy or to help direct patients to watchful waiting in lieu of surgery. The study also found that the GEC result was frequently misinterpreted when positive (suspicious) by the referring endocrinologists and even among our own surgeons, leading to overtreatment. Therefore, we set out to validate the findings prospectively by analyzing the pattern of use, effect, and outcome of molecular profiling of thyroid nodules in patients being surgically evaluated at The Johns Hopkins Hospital. We found that molecular profiling did not significantly affect the surgical decision-making process in 92.1% (129 of 140) of our surgical patients. When only examining patients with indeterminate nodules (atypia of undetermined significance, suspicious for follicular neoplasm, or suspicious for Hürthle cell neoplasm), we found that molecular profiling did not significantly affect the surgical decision-making process in 89.9% (98 of 109) of patients. This finding clearly demonstrates that these tests are being overused in patients for whom the results would not change the surgical management. Moreover, routine sampling at the time of initial FNA biopsy and selective repeated FNA biopsy for molecular profiling are associated with extra time and effort by the physician and patients, as well as additional cost. This information serves to help standardize how we should approach patients with thyroid nodules in an era where there is an increased demand for accountability when delivering health care.
Our results showed that in 50.7% (71 of 140) of the cases molecular profiling was not used for its intended purpose, namely, to differentiate histologically benign from malignant indeterminate or suspicious thyroid nodules. An example is molecular testing used in patients with malignant FNA findings, presumably to advocate for a total thyroidectomy or prophylactic central neck dissection. Three patients had nodules suspicious for malignancy on cytology assessment and underwent GEC testing, which is intended to rule out cancer. In addition, there were 21 patients with benign or insufficient cytology findings who had inappropriate ordering of molecular testing as well. Among these cases when the molecular test result affected the surgical decision making, all of the patients were overtreated. The latter scenarios indicate that molecular tests can actually hinder the surgical decision-making process when not clinically indicated.
Thus far, no molecular test can replace the cytopathological diagnosis in the management of thyroid nodules, which is owing to the fact that no single molecular marker has shown clinically acceptable negative predictive values or positive predictive values to allow its use as a sole diagnostic tool in evaluating thyroid nodules. Therefore, molecular profiling is just one of many adjuncts that may be undertaken in the evaluation of a thyroid nodule. Current evidence suggests that GEC testing may be beneficial in minimizing unnecessary surgical procedures in a subset of patients with an indeterminate (BSRTC classification III or IV) thyroid nodule.1,9 However, as shown in this study, physicians must continue to use the diagnostic tools available to them when evaluating a thyroid nodule, instead of solely relying on a molecular test or reflexively obtaining one. To avoid unnecessary testing, it may be prudent for surgeons to be consulted before the initiation of molecular profiling. Nonetheless, physical examination, imaging, and FNA cytology evaluation, coupled with the treating physician’s judgment and patient preference, should guide the management of indeterminate and suspicious nodules. As such, physicians should consider using the GEC test as an ancillary diagnostic tool in patients with asymptomatic thyroid nodules between 1 and 4 cm in whom an adequate FNA biopsy has been performed yielding a cytological diagnosis of atypia of undetermined significance, suspicious for follicular neoplasm, or suspicious for Hürthle cell neoplasm, with minimal clinical suspicion for a malignant tumor. Even as a rule-out test, GEC testing has a false-negative rate of up to 15% (possibly higher at some institutions), so continued follow-up assessment is crucial.9,11 Conversely, a suspicious GEC result should not be interpreted as cancer. If surgery is elected, the need for thyroid lobectomy only or a total thyroidectomy should depend on the clinical scenario and patient preference. For patients with an indeterminate cytological diagnosis (BSRTC classification III, IV, or V), physicians may consider performing gene mutational analysis because of its reported higher positive predictive value only if there is a suspicious result from a rule-out test. Yet, the combined cost of GEC testing followed by SMP tests could be prohibitive at the present rate. For patients with cytology results suspicious for malignancy (BSRTC classification V), gene mutation markers or SMP testing would be reasonable given the high cancer prevalence, yet only if the patient and treatment team are not already committed to a total thyroidectomy based on the clinical findings.
This study has several limitations. Because all of the molecular testing was initiated by the referring physicians, selection bias is likely (eg, a greater suspicion about the thyroid nodule). Unlike other molecular tests, the GEC test is offered at The Johns Hopkins Hospital. Because endocrinologists and internists are the first-line caregivers, they were more likely to obtain this test for patients before referral. It is also highly likely that those patients having negative or benign molecular testing results who did not have any other considerations for surgery were less likely to be referred for surgical management, thus biasing our study population. Although the surgeons in this study universally agreed on the surgical algorithm into which the markers were incorporated, the study is biased by our institutional management philosophy, and the results may not be universally applicable because not all institutions share our approach for the management of thyroid nodules. Also, the potential variance of the positive predictive value and negative predictive value of these molecular tests between institutions might affect physician reliance on test results for determining the need for and extent of surgery. Therefore, the indication, effect, and value of such testing need to be determined by individual institutions and treatment teams.
Overall, physical examination, imaging, and FNA cytology evaluation, coupled with the treating physician’s judgment and patient preference, should guide the management of indeterminate and suspicious nodules. Commercially available molecular testing of thyroid nodules did not significantly affect the surgical decision-making process in this study. Among patients whose treatment was altered based on these markers, there was evidence of overtreatment.
Diagnostic molecular testing should not be used reflexively or in isolation to drive clinical decision making in patients with cytologically indeterminate or suspicious thyroid nodules. These tests should only be considered once patient, clinical, radiological, and cytological factors are weighed in a balanced manner and it is determined that a molecular test result would appropriately alter the treatment plan.
Accepted for Publication: March 19, 2016.
Corresponding Author: Ralph P. Tufano, MD, MBA, Division of Head and Neck Endocrine Surgery, Department of Otolaryngology–Head and Neck Surgery, The Johns Hopkins University School of Medicine, 601 N Caroline St, Sixth Floor (Room 6242), Baltimore, MD 21287 (firstname.lastname@example.org).
Published Online: May 19, 2016. doi:10.1001/jamaoto.2016.0850.
Author Contributions: Drs Noureldine and Tufano had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Noureldine, Aragon Han, Olson, Schneider, Prescott, Mathur, Zeiger, Tufano.
Acquisition, analysis, or interpretation of data: Noureldine, Najafian, Aragon Han, Olson, Genther, Schneider, Agrawal, Mathur, Zeiger, Tufano.
Drafting of the manuscript: Noureldine, Najafian, Zeiger.
Critical revision of the manuscript for important intellectual content: Aragon Han, Olson, Genther, Schneider, Prescott, Agrawal, Mathur, Zeiger, Tufano.
Statistical analysis: Noureldine, Najafian, Genther.
Administrative, technical, or material support: Najafian, Aragon Han, Olson, Prescott, Agrawal, Tufano.
Study supervision: Noureldine, Olson, Schneider, Prescott, Agrawal, Mathur, Zeiger, Tufano.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest, and none were reported.
Previous Presentation: This study was presented at The American Association of Endocrine Surgeons 36th Annual Meeting; May 17, 2015; Nashville, Tennessee.