Abstract
Molecular diagnostics increasingly direct the management of thyroid nodules with an indeterminate cytologic diagnosis. This study was undertaken to correlate cytomorphologic features with the molecular profiles in an effort to identify features predictive of molecular aberrations.
189 thyroid nodules with an indeterminate thyroid cytology diagnosis (atypia of undetermined significance, suspicious for follicular lesion, and suspicious for malignancy) with an adequate sample submitted for targeted mutation detection by PCR or next generation sequencing (NGS) were assessed semi-quantitatively for following cytomorphologic parameters: (cellularity, Hurthle cell changes, microfollicles, nuclear elongation, nuclear grooves, nuclear enlargement, nuclear atypia, extent of atypia and colloid). Based on this evaluation, a cumulative cytomorphologic score (CCS) and a more simplified overall atypia score (OAS) were assigned to each case. Associations among mutational status and each of the aforementioned parameters, CCS and OAS scores were determined. Of the 189 nodules with indeterminate cytology, 63 (33.3%) harbored at least one mutation. RAS and BRAF were the most common mutations, found in 34 (18.0%) and 13 (6.9%) cases respectively. Both CCS and OAS were highly associated with the presence of all mutations (p<0.0001) and with the presence of BRAF and RAS mutations in particular (all p <0.01).
Semi-quantitative assessment of various cytomorphologic features in indeterminate thyroid cytology cases showed a strong association of higher OAS and CCS scores and incidence of BRAF and RAS mutations. Utilizing a more objective approach to thyroid cytology can potentially decrease the overall number of indeterminate diagnoses leading to fewer repeat procedures and unnecessary surgeries.
Keywords: Thyroid nodules, thyroid FNA, atypia of undetermined significance, follicular lesions of undetermined significance, suspicious for malignancy, mutational profile
Introduction
Thyroid nodules are common with a reported prevalence as high as 68% when detected by ultrasound [1,2]. Current guidelines recommend fine needle aspiration (FNA) cytology in clinically concerning nodules or nodules greater than or equal to 1 cm [1,3]. In the US, reporting of thyroid cytology is primarily based on the Bethesda System for Reporting Thyroid Cytology (TBSRTC). TBSRTC includes six diagnostic categories. Each category is stratified by risk of malignancy and recommended clinical management [4]. The cytomorphologic features described in the TBSRTC have been used by the cytopathologists for over a decade; however, wide variation in reporting of TBSRTC categories, among institutions suggests inconsistent application of these criteria. The most marked discrepancies occur with indeterminate diagnostic categories: atypia of undetermined significance (AUS), suspicious for follicular neoplasm (FN) and suspicious for malignancy (SM). Indeterminate categories carry the highest degree of uncertainty in terms of diagnosis and risk of malignancy thereby leading to repeat procedures and unwarranted surgeries. These categories constitute 9.9 to 38.2% of all cytologic TBSRTC categories in general practice [5–10]. In order to tackle the relatively high rate of indeterminate diagnosis, commercial, non-commercial, and institutional molecular testing are used to augment cytologic diagnosis [11]. Such testing provides an opportunity to further refine the conventional cytomorphologic criteria of thyroid lesions.
The purpose of this study was to semi-quantitatively analyze the cytomorphologic features associated with indeterminate cytology using two different scoring schemes, looking for phenotypic patterns that might be predictive of specific molecular alterations. Associations between each of the scoring schemes and a mutation panel were used to assess scoring scheme performance. Such a system might provide more objective process in making indeterminate thyroid cytology diagnoses.
Materials and Methods
Study Cohort
After receiving IRB approval, a retrospective review of all thyroid FNA performed between 2013-2015 was conducted. During this period, 1172 nodules were sampled. Nodules with indeterminate TBSRTC categories (AUS, FN, and SM) which also had molecular results were selected for further study. The archived cytology slides from selected cases were retrieved for morphologic evaluation and in cases where a subsequent resection was performed; histologic diagnosis was correlated to the cytology and molecular results.
Molecular Testing
In most of the cases, FNA was performed with 4 passes and rapid onsite adequacy evaluation. Direct smears were made from passes # 1,2 and 4 while needle rinses from these passes and all of pass #3 were place in RNA/DNA stabilization reagent (Roche) and stored at 4°C. Samples with indeterminate diagnosis, were submitted to the University of Pittsburgh Medical Center molecular diagnostic lab for targeted mutation detection by PCR or by next generation sequencing as part of ongoing clinical care, either as a result of reflex testing or by specific clinician request. From January of 2013 to September of 2013 (63 nodules), seven-gene mutational testing (MT) was performed that utilized real-time LightCycler polymerase chain reaction and fluorescence melting curve analysis to detect possible mutations (BRAF, NRAS codon 61, HRAS codon 61, KRAS codons 12 and 13) and single-step real-time reverse transcription real-time polymerase chain reaction to amplify the fusion points of the rearrangements (RET/PTC1, RET/PTC3, and PAX8/PPARG). Subsequently, ThyroSeq® NGS-based versions 1 and 2 were used as they were made available from the reference laboratory [13–14]. Thyroseq® v1 included NGS for 284 mutations in 12 key thyroid cancer related genes including BRAF, AKT1, CTNNB1, GNAS, NRAS, HRAS, KRAS, PIK3CA, PTEN, RET, TP53, and TSHR, and detection of chromosomal rearrangements RET/PTC1, RET/PTC3, and PAX8/PPARG. Thyroseq® v2 included the above point mutations plus EIF1AX, TERT, and 42 gene fusions involving the following genes: RET, PPARG, NTRK1, NTRK3, ALK, BRAF, and IGF2BP3.
Semi-quantitative evaluation of cytomorphologic features
The FNA material was reviewed simultaneously by two pathologists on a double headed microscope. Pathologists were blinded to the mutational status, cytologic diagnosis, or histologic diagnosis of any subsequent resection. The minimal cellularity criteria used for adequacy was presence of 6 groups of follicular cells each with at least 10 cells as defined in the TBSRTC. Nine cytomorphologic parameters were semi-quantitatively scored for each case as follows: cellularity (1=hypocellular; 2=moderately cellular; 3= hypercellular), Hurthle cell changes (0=no Hurthle cells; 1=focal Hurthle cells; 2=Abundant/diffuse Hurthle cells), microfollicles (0=no microfollicles; 1=occasional; 2=predominant), nuclear atypia which is a constellation of features including, nuclear membrane irregularity, nuclear overlapping, fine chromatin and conspicuous nucleoli (0=no atypia; 1=mild; 2=moderate; 3:marked), nuclear enlargement (3-4 times of RBC) (0=none; 1=focal; 2=multifocal), nuclear elongation (0=none; 1=focal; 2=multifocal), nuclear grooves (0=none;1=focal; 2=multifocal), extent of atypia (0=no atypia; 1=focal; 2=multifocal; 3=diffuse) and colloid (0=no colloid; −1=scant; −2=moderate; - 3=abundant) [15].
The sum of all these scores generated a cumulative cytomorphologic score (CCS) with a range from −3 to 18 (Table 1). In addition to individual cytomorphologic feature evaluation, each case was assigned a separate more simplified simplified overall atypia score (OAS) based on pathologist’s level of suspicion (1= less likely neoplastic; 2=moderate; 3= most likely neoplastic). OAS can be best described as cytopathologists level of suspicion, taking into consideration all cytomorphological features, for the likelihood of neoplasia. We used this simplified score which is more representative and easier to implement in routine practice of reporting thyroid cytology and compared its performance to more objective and somewhat complex assessment by CCS.
Table 1.
Scoring scheme of the cytomorphologic parameters determining the cumulative cytomorphologic score (CCS)
| 0 | 1 | 2 | 3 | |
|---|---|---|---|---|
| Cellularity | Hypocellular | Moderately cellular | Hypercellular | |
| Hurthle cell changes | None | Focal | Abundant/diffuse | |
| Microfollicle | None | Occasional | Predominant | |
| Nuclear atypia | None | Mild | Moderate | Marked |
| Extent of atypia | None | Focal | Multifocal | |
| Nuclear enlargement | None | Focal | Multifocal | |
| Nuclear elongation | None | Focal | Multifocal | |
| Nuclear grooves | None | Focal | Multifocal | |
| 0 | −1 | −2 | −3 | |
| Colloid | None | Scant | Moderate | Abundant |
The sum of the scores on each row determines the CCS which ranges from −3 to 18.
Statistical Methods
Demographic and pathological characteristics were summarized and compared by mutation status (any, BRAF, RAS) compared to no mutation (or no BRAF or RAS, respectively) using Chi-Squared and Fisher’s Exact tests for categorical data and t-tests and Wilcoxon Rank Sum tests for continuous data as appropriate. Potential cut-offs for the CCS were determined using receiver operating characteristic (ROC) curves and the corresponding area under the curve (AUC). Cut-offs were chosen to maximize sensitivity while maintaining specificity of 50% (CCS <7/7+) and maximize AUC (both sensitivity and specificity; CCS <10/10+) for comparisons by mutation status (any vs. none). Point estimates and exact 95% confidence intervals (Cl) for the sensitivity and specificity of these cutoff scores are presented comparing the presence of any mutation, BRAF and RAS to no mutation, no BRAF or no RAS mutations, respectively. Since the SM category nodules were associated with higher cytomorphologic feature scores, a sensitivity analysis was conducted, excluding the 17 SM category nodules from the analysis. Data were analyzed using SAS 9.3 (Cary, NC) and p-values less than 0.05 were considered statistically significant.
Results
The 1172 distinct thyroid nodules aspirated during this study period were categorized as follows according to TBSRTC. Indeterminate cytologic diagnosis was rendered in 250 (21.3%) cases, 184 (15.6%) were AUS, 36 (3.1%) SFN/FN and 30 (2.6%) were SM. Of these, 228 (19.5%) were submitted for molecular studies. Reportable molecular results were available on 189 aspirates from 169 patients who were selected for this study. All the 189 aspirates also met the criteria for adequate cellularity. 134 of the patients were female and 35 male.
The FNA diagnosis was AUS in 146 cases, SFN in 26 cases and SM in 17 cases. Mutations were identified in 63 cases with RAS identified as the most common (n=34, 18.0%) and BRAF the second most common mutation (n=13; 6.9%). Three nodules harbored more than one mutation (Table 2). 78 nodules were surgically excised.
Table 2.
Cytology, mutation status and type of mutation (N=189).
| N | % | |
|---|---|---|
| Cytology | ||
| AUS | 146 | 77.3 |
| SFN/FN | 26 | 13.7 |
| SM | 17 | 9.0 |
| Mutation | ||
| No | 126 | 66.7 |
| Yes | 63 | 33.3 |
| Mutation Type | ||
| ALK | 1 | 1.6 |
| BRAF | 12 | 19.1 |
| BRAF, TERT, PIK3CA, AKT1 | 1 | 1.6 |
| EIFIAX | 1 | 1.6 |
| GNAS | 2 | 3.2 |
| HRAS | 6 | 9.5 |
| HRAS, RET | 1 | 1.6 |
| KRAS | 9 | 14.3 |
| NRAS | 17 | 27.0 |
| NRAS, TSHR | 1 | 1.6 |
| P53 | 4 | 6.4 |
| PAX8/PPARg | 2 | 3.2 |
| PTEN | 2 | 3.2 |
| TSHR | 4 | 6.4 |
Correlation of individual cytomorphologic features with mutational status
Table 3 shows the prevalence of each of 9 individual cytomorphologic features score by presence or absence of mutation. Extent of atypia i.e. focal, multifocal or diffuse (p<0.0001), nuclear enlargement (p<0.0001), nuclear elongation (p<0.0001), nuclear atypia (p=0.0007), nuclear grooves (p=0.0005) and cellularity (p=0.01) were significantly different between mutation positive with mutation negative nodules. No differences by mutation status were observed for Hurthle cell change, microfollicles or absence of colloid.
Table 3:
Correlation of individual cytomorphologic parameters and overall atypia score with mutational status
| No Mutatio n (N=126) | Any Mutation (N=63) | No BRAF (N=176) | BRAF (N=13) | No RAS (N=155) | RAS (N=34) | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| N | % | N | % | p-value | N | % | N | % | p-value | N | % | N | % | p-value | |
| 0 | 29 | 23.0 | 14 | 22.2 | 38 | 21.6 | 5 | 38.5 | 36 | 23.2 | 7 | 20.6 | |||
| 1 | 68 | 54.0 | 27 | 42.9 | 92 | 52.3 | 3 | 23.1 | 80 | 51.6 | 15 | 44.1 | |||
| 2 | 29 | 23.0 | 22 | 34.9 | 46 | 26.1 | 5 | 38.5 | 39 | 25.2 | 12 | 35.3 | |||
| Cellularity (1-3) | 0.01 | 0.44 | 0.001 | ||||||||||||
| 1 | 21 | 16.7 | 4 | 6.4 | 25 | 14.2 | 0 | 0.0 | 24 | 15.5 | 1 | 2.9 | |||
| 2 | 82 | 65.1 | 37 | 58.7 | 109 | 61.9 | 10 | 76.9 | 102 | 65.8 | 17 | 50.0 | |||
| 3 | 23 | 18.3 | 22 | 34.9 | 42 | 23.9 | 3 | 23.1 | 29 | 18.7 | 16 | 47.1 | |||
| Microfollicles (0-3) | 0.24 | 0.86 | 0.002 | ||||||||||||
| 0 | 42 | 33.3 | 20 | 31.8 | 57 | 32.4 | 5 | 38.5 | 56 | 36.1 | 6 | 17.7 | |||
| 1 | 59 | 46.8 | 23 | 36.5 | 77 | 43.8 | 5 | 38.5 | 69 | 44.5 | 13 | 38.2 | |||
| 2 | 20 | 15.9 | 14 | 22.2 | 31 | 17.6 | 3 | 23.1 | 25 | 16.1 | 9 | 26.5 | |||
| 3 | 5 | 4.0 | 6 | 9.5 | 11 | 6.3 | 0 | 0.0 | 5 | 3.2 | 6 | 17.7 | |||
| Nuclear Atypia (0-3) | 0.0007 | <0.0001 | 0.26 | ||||||||||||
| 0 | 5 | 4.0 | 0 | 0.0 | 5 | 2.8 | 0 | 0.0 | 5 | 3.2 | 0 | 0.0 | |||
| 1 | 95 | 75.4 | 35 | 55.6 | 128 | 72.7 | 2 | 15.4 | 110 | 71.0 | 20 | 58.8 | |||
| 2 | 23 | 18.3 | 19 | 30.2 | 37 | 21.0 | 5 | 38.5 | 31 | 20.0 | 11 | 32.4 | |||
| 3 | 3 | 2.4 | 9 | 14.3 | 6 | 3.4 | 6 | 46.2 | 9 | 5.8 | 3 | 8.8 | |||
| Nuclear Enlargement | <0.0001 | <0.0001 | 0.09 | ||||||||||||
| 0 | 16 | 12.7 | 0 | 0.0 | 16 | 9.1 | 0 | 0.0 | 16 | 10.3 | 0 | 0.0 | |||
| 1 | 103 | 81.8 | 46 | 73.0 | 146 | 83.0 | 3 | 23.1 | 121 | 78.1 | 28 | 82.4 | |||
| 2 | 7 | 5.6 | 17 | 27.0 | 14 | 8.0 | 10 | 76.9 | 18 | 11.6 | 6 | 17.7 | |||
| Nuclear elongation | <0.0001 | <0.0001 | 0.01 | ||||||||||||
| 0 | 90 | 71.4 | 28 | 44.4 | 116 | 65.9 | 2 | 15.4 | 104 | 67.1 | 14 | 41.2 | |||
| 1 | 36 | 28.6 | 29 | 46.0 | 58 | 33.0 | 7 | 53.9 | 47 | 30.3 | 18 | 52.9 | |||
| 2 | 0 | 0.0 | 6 | 9.5 | 2 | 1.1 | 4 | 30.8 | 4 | 2.6 | 2 | 5.9 | |||
| Nuclear grooves | 0.0005 | <0.0001 | 0.26 | ||||||||||||
| 0 | 108 | 85.7 | 42 | 66.7 | 148 | 84.1 | 2 | 15.4 | 126 | 81.3 | 24 | 70.6 | |||
| 1 | 18 | 14.3 | 16 | 25.4 | 27 | 15.3 | 7 | 53.9 | 25 | 16.1 | 9 | 26.5 | |||
| 2 | 0 | 0.0 | 5 | 7.9 | 1 | 0.6 | 4 | 30.8 | 4 | 2.6 | 1 | 2.9 | |||
| Focal/Diffuse (extent of) atypia | <0.0001 | <0.0001 | 0.003 | ||||||||||||
| 0 | 8 | 6.4 | 2 | 3.2 | 10 | 5.7 | 0 | 0.0 | 8 | 5.2 | 2 | 5.9 | |||
| 1 | 88 | 69.8 | 24 | 38.1 | 110 | 62.5 | 2 | 15.4 | 101 | 65.2 | 11 | 32.4 | |||
| 2 | 25 | 19.8 | 23 | 36.5 | 44 | 25.0 | 4 | 30.8 | 33 | 21.3 | 15 | 44.1 | |||
| 3 | 5 | 4.0 | 14 | 22.2 | 12 | 6.8 | 7 | 53.9 | 13 | 8.4 | 6 | 17.7 | |||
| Colloid | 0.19 | 0.13 | 0.85 | ||||||||||||
| 0 | 51 | 40.5 | 31 | 49.2 | 73 | 41.5 | 9 | 69.2 | 66 | 42.6 | 16 | 47.1 | |||
| −1 | 43 | 34.1 | 13 | 20.6 | 52 | 29.6 | 4 | 30.8 | 48 | 31.0 | 8 | 23.5 | |||
| −2 | 21 | 16.7 | 15 | 23.8 | 36 | 20.5 | 0 | 0.0 | 29 | 18.7 | 7 | 20.6 | |||
| −3 | 11 | 8.7 | 4 | 6.4 | 15 | 8.5 | 0 | 0.0 | 12 | 7.7 | 3 | 8.8 | |||
| No Mutation (N=126) | Any Mutation (N=63) | No BRAF (N=176) | BRAF (N=13) | No RAS (N=155) | RAS (N=34) | ||||||||||
| N | % | N | % | p-value | N | % | N | % | p-value | N | % | N | % | p-value | |
| Overall atypia Score (0-3) | <0.0001 | <0.0001 | 0.0002 | ||||||||||||
| 1 | 67 | 53.2 | 15 | 23.8 | 82 | 46.6 | 0 | 0.0 | 75 | 48.4 | 7 | 20.6 | |||
| 2 | 37 | 29.4 | 17 | 27.0 | 52 | 29.6 | 2 | 15.4 | 46 | 29.7 | 8 | 23.5 | |||
| 3 | 22 | 17.5 | 31 | 49.2 | 42 | 23.9 | 11 | 84.6 | 34 | 21.9 | 19 | 55.9 | |||
Further analysis of the individual cytomorphologic features by specific mutation type was performed for RAS and BRAF. BRAF mutation was associated with the same general pattern of cytopathologic feature score significance as the overall group, except, cellularity score was no longer significant (p=0.44). In addition to the pattern shown for the overall group, the presence of RAS mutations also correlated significantly with microfollicles (p=0.002).
Since the suspicious for malignancy category nodules were associated with higher cytomorphologic feature scores, statistical analysis was also performed after excluding the 17 SM category nodules from the analysis. This did not impact cytomorphologic score statistical significance by mutation status except for “cellularity” feature, which did not retain its significance (p=0.09).
Correlation of CCS with mutational status
The CCS scores were widely distributed, ranging from −2 to 18 (Table 4). The 63 mutation positive nodules had a median CCS of 10 compared to a median score of 6 in the 126 mutation negative nodules (p<0.0001). BRAF positive nodules exhibited a higher median CCS compared to RAS positive nodules, 12 versus 10.5.
Table 4:
Performance of Cumulative Cytomorphologic Score (CCS) and atypia score in determining the mutation status.
| No Mutation (N=126) | Any Mutation (N=63) | No BRAF (N=176) | BRAF (N=13) | No RAS (N=155) | RAS (N=34) | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| N | Median (Range) | N | Median (Range) | p-value | N | Median (Range) | N | Median (Range) | p-value | N | Median (Range) | N | Median (Range) | p-value | |
| CCS | 126 | 6 (−2-16) | 63 | 10 (2-18) | <0.0001 | 176 | 7 (−2-16) | 13 | 12 (8-18) | <0.0001 | 155 | 7 (−2-18) | 34 | 10.5 (3-15) | 0.0003 |
| N | % | N | % | p-value | N | % | p-value | N | % | p-value | |||||
| CCS Cut-off 1 | 0.0009 | 0.0007 | 0.005 | ||||||||||||
| <7 | 64 | 50.8 | 16 | 25.4 | 80 | 45.5 | 0 | 0.0 | 73 | 47.1 | 7 | 20.6 | |||
| 7+ | 62 | 49.2 | 47 | 74.6 | 96 | 54.6 | 13 | 100.0 | 82 | 52.9 | 27 | 79.4 | |||
| CCS Cut-off 2 | <0.0001 | <0.0001 | <0.0001 | ||||||||||||
| <10 | 107 | 84.9 | 31 | 49.2 | 136 | 77.3 | 2 | 15.4 | 124 | 80.0 | 14 | 41.2 | |||
| 10+ | 19 | 15.1 | 32 | 50.8 | 40 | 22.7 | 11 | 84.6 | 31 | 20.0 | 20 | 58.8 | |||
| Atypia Score | <0.0001 | <0.0001 | <0.0001 | ||||||||||||
| 1 or 2 | 104 | 82.5 | 32 | 50.8 | 134 | 76.1 | 2 | 15.4 | 121 | 78.1 | 15 | 44.1 | |||
| 3 | 22 | 17.5 | 31 | 49.2 | 42 | 23.9 | 11 | 84.6 | 34 | 21.9 | 19 | 55.9 | |||
CCS cut-offs of 7 and 10 were calculated based on ROC analysis defining maximum sensitivity and specificity, respectively. 80 nodules (16 with mutations and 64 without mutations) fell below a CCS cut-off of 7. 43.1% (47 of 109) nodules with CCS of 7 or more were positive for mutation compared to 20% (16 of 80) with CCS <7 (p=0.0009) (figure 1). At this cut-off, the sensitivity and specificity for detecting mutation was 74.6% (95% CI: 62.1,84.7) and 50.8% (95% CI: 41.7, 59.8) respectively; Using a cut-off score of 10, 22.5% (31 of 138) cases with CCS <10 were positive for mutations and 62.7% (32 of the 51) with a CCS score of 10 or more were positive for mutation (p=<0.0001). At this cut-off, the sensitivity and specificity for detecting mutation was 50.8% (95% CI: 37.9, 63.6) and 84.9% (95% CI: 77.5, 90.7), respectively. CCS cut off ≥10 or higher performed slightly better than CCS cut off >7 in accurately predicting mutation (positive likelihood ratio 3.36 vs 1.5).
Figure 1-.
Distribution of mutations using cumulative cytomorphologic score cut-offs of 7 and 10.
For detection of BRAF mutation only, the sensitivity and specificity of scores equal to or above 7 was 100% (95% CI; 75.3, 100.0) and 45.5% (95% CI: 38.0, 53.1) respectively and for scores equal to and above 10 the sensitivity and specificity was 84.6% (95% CI: 54.6, 98.1) and 77.3% (95% CI: 70.4, 83.2), respectively. For RAS mutations, the sensitivity and specificity of scores equal to or above 7 was 79.4% (95% CI: 62.1, 91.3) and 47.1% (95% CI: 39.0, 55.3) respectively and for scores equal to and above 10 the sensitivity and specificity was 58.8% (95% CI: 40.7, 75.4) and 80.0% (95% CI: 72.8, 86.0), respectively.
Correlation of OAS with mutational status
The OAS was also highly associated with the mutation status (p<0.0001, Table 3). In the nodules where mutations were detected, 49.2% of the overall population had an atypia score of 3, whereas 84.6% BRAF mutation positive and 55.9% RAS mutation positive nodules had an atypia score of 3. Conversely, the majority of cases (53.2%) without mutation(s) had atypia score of 1.
The correlation of mutation status and the OAS was still statistically significant after excluding nodules in the SM category (data not shown).
If the status of mutation is considered as the gold standard and OAS of 3 to define a positive test, sensitivity and specificity of OAS for predicting mutational status were 49.2% (95% CI: 36.4, 62.1) and 82.5% (95% CI: 74.8, 88.7), respectively.
Correlation of histologic diagnosis with CCS and OAS
Follow up resection data were available for 78 of the 189 nodules. 20 were non-neoplastic (adenomatoid nodule, nodular hyperplasia, Hashimoto thyroiditis) (median CCS score=6, range 2-15); 24 were categorized as benign neoplasm (follicular adenoma) (median CCS score=9.5, range 4-13) and 34 were categorized malignant neoplasm (4 follicular carcinomas, 28 papillary thyroid carcinomas, 2 medullary thyroid carcinomas) (median CCS score=11, range: 4-18). CCS was significantly higher in malignant neoplasm group compared to the non-neoplastic (p=0.0004) and benign (p=0.02) groups. Similarly, the OAS showed a statistically significant difference among the three groups (p=0.0002). To assess the impact of noninvasive follicular thyroid neoplasm with papillary-like nuclear features (NIFTP), a new entity introduced in 2016 for borderline tumors (16), all surgical specimens in this cohort were re-reviewed and 11 cases reclassified as NIFTP. Previous histologic diagnosis of these cases was as follows: 6 PTC, 4 adenomas and 1 nodular hyperplasia. However, the rate of malignancy is in this study is determined by the original histologic classification. By cytology the NIFTP cases were categorized as follows: 6 AUS, 4 SFN and 1SM. 8 of the 11 cases (73%) were positive for mutations with 5 RAS mutations, one P53, one BRAF K601E and one PAX8/PPARg translocation. The median CSS of NIFTP cases was 9 (mean:9.6, range 5-13), OAS was 3 in 5 cases, 2 in 5 cases and 1 in one case. These scores are more closer to malignant than benign neoplasms.
Discussion
TBSRTC has standardized classification and management of thyroid nodules; however, cytomorphologic evaluation and interpretation remains subjective [6]. Reviewer bias of indeterminate nodules (AUS and FN) is concerning. Molecular testing offers an objective single result for determining malignancy risk and triage for diagnostic surgery for these intermediate groups [11–15,17]. At the same time, successful molecular testing often requires additional sampling and has added cost. Subclassification of the indeterminate cytologic categories by cytologic features provides another potential tool to determine malignancy risk.
This study semi-quantitatively evaluated 9 cytomorphologic parameters. Among the individual cytomorphologic features analyzed, cellularity, nuclear atypia, nuclear enlargement, elongation, grooving and extent of atypia, showed statistically significant differences, but couldn’t individually distinguish across mutation/rearrangement status. A CCS cut off of 10 provided the greatest discrimination for mutation detection based on ROC analysis. CCS ≥10 had high sensitivity and specificity for detecting mutations in general and even higher sensitivity (84.6%) and specificity (77.3%) for BRAFmutation, in particular (figure 2). A high degree of correlation was also found for atypia score 3 and mutation presence (49.2%) versus atypia score 1 and absence of mutation (53.2%).
Figure 2:
Representative photomicrographs of thyroid FNA with AUS diagnosis and low and high CCS (Diff Quik, 40x); A, CCS of 6 with no mutation detected, nodular hyperplasia on histology; B, CCS of 10 with NRAS mutation, follicular adenoma on histology; C, CCS of 11 with BRAF mutation, papillary carcinoma on histology.
Assessing the sensitivity and specificity of OAS along with CSS for detection of molecular aberrations, enabled us to compare the performance of both subjective and objective approaches in parallel. OAS appears to be more sensitive (82.5% vs. 50.8%) while CCS is more specific (84.9% vs. 49.2%) in predicting molecular aberration status. If we hypothetically consider CCS of 10 or OAS of 3 as cut-off for recommendation for surgery, equivalent to SFN or SM according to TBSRTC, it would result in 26 AUS cases to be upgraded with 2.2% reduction in AUS cases and 3.2% decrease in the rate of malignancy in the remaining cases (5% vs. 8.2%) in this study cohort. A significantly higher rate of malignancy (23%) in the upgraded AUS cases reflects the usefulness of these scoring schemes in capturing high-risk nodules.
In addition to cytology-molecular correlation, histologic correlation was also made where histologic diagnosis was available and showed both CCS and OAS scores significantly higher in malignant neoplasm. A total of 5 nodules with final histologic diagnosis of malignancy, failed to show any genetic aberration.
Actually two of these nodules were found to be medullary thyroid carcinoma and RET gene was not included in the initial version of assay therefore part of this discrepancy can be attributed to tumors having mutations not targeted by the assay. Another reason could be inadequate sampling, a recognized pitfall in cytology, particularly the use of dedicated pass for molecular studies that is not subjected to rapid on-site cytologic evaluation for adequacy. More recent assays which retrieve material from a Diff Quik stained slide known to contain diagnostic material improves the yield [18].
A few other studies have stratified malignancy risk using cytologic morphologic qualifiers. Renshaw found the risk of malignancy was significantly different across subclasses of atypical follicular cell category (rule out papillary carcinoma vs. rule out Hurthle cell neoplasm categories) [9]. Wu et al subclassified AUS as AUS-PTC and AUS-FN and found that it carried a higher risk of malignancy and neoplasm respectively than their other 2 described AUS subclasses [19].
Sherestha et al and Johnson et. al. used cytologic versus architectural atypia for risk stratification of AUS/FLUS group, showing higher malignancy rate in cases with cytologic atypia [20, 21]. Correlation of cytologic findings with molecular results has also been reported by a few others [22, 23]. Beca et al. showed AUS subclassification on the basis of cytologic atypia alone or cytoarchitectural atypia correlated with suspicious Afirma GEC results [24]. Bellevicine et al demonstrated BRAF and RAS mutations are associated with different AUS qualifiers (cytologic atypia, architectural atypia and Hurthle cell changes) and therefore carrying different risks of malignancy [25]. RAS mutations which are known to be associated with follicular pattern thyroid neoplasms [26] were also found to exhibit statistically significant association with microfollicle score in this study. A study by Rossi et al had shown high incidence of BRAF mutation in thyroid cytology specimens with at least focal plump cells with abundant eosinophilic cytoplasm [27], in our study, 4 out of 13 BRAF positive cases had a high Hurthle cell score (although statistically not significant). BRAF positive cases have been shown to be associated with higher incidence of tall cell variant, an aggressive type of PTC [28], however based on cytology-histology correlation; no case in our study was classified as tall cell variant.
The rationale for using molecular data as a reference to compare with morphologic features is that, currently, it is probably the most objective methods to access neoplastic nature of the lesion, which is also supported by strong evidence of association of molecular aberrations and malignant histologic diagnosis in this study. Furthermore, such an approach can also help optimally triage the samples; i.e., whether the patient would benefit from ancillary mutational analysis that can provide crucial information for clinical decision making. This study does not advocate a laborious scoring system for routine use in thyroid cytology sign outs but emphasizes the importance of an objective approach backed by molecular data, which could reduce the number of indeterminate cases and avoid unnecessary work up and overtreatment.
This study included cases from a single academic center with a relatively modest number of indeterminate cytology cases that were subjected to molecular testing by two different test platforms (PCR based MT and NGS based Thyroseq®V1 and V2) which could be considered as potential limitations. Although the analytic performance of both platforms is comparable [13], Thyroseq®V1 and V2 has higher clinical sensitivity because of its ability to detect more mutations [11–14]. However, it is interesting to note that most of the additional mutations (TSHR, PTEN, ALK, GNAS and P53) detected by Thyroseq® in our study were associated with benign nodules. Only 1 case in this study with EIF1AX mutation specific to Thyroseq®V2 turned out to be malignant. Molecular testing in this study was only approved for cases with indeterminate cytologic diagnosis and therefore was not performed on benign TBSRTC category which could have provided further insight into the ability of CCS and OAS to differentiate between mutation negative benign versus mutation positive benign versus indeterminate benign cases with or without mutations.
In summary, the prevalence of mutation(s) in indeterminate TBSRTC category cases was associated with higher semi-quantitatively determined cytomorphologic scores. These data suggest the implementation of more objective scoring method may be used to refine and more precisely categorize indeterminate thyroid FNAs based on TBSRTC.
Highlights:
RAS mutations are the most common aberrations in indeterminate thyroid nodules
Semi-quantitative cytomorphologic assessment correlates with mutational status
Indeterminate thyroid cytology cases can be reduced with more objective approach
Footnotes
Publisher's Disclaimer: This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Disclosures and acknowledgements
The authors report no conflicts of interest. No funding or support is associated with this study.
This study has been presented as an abstract at the USCAP annual meeting in 2017.
References
- 1.Haugen BR, Alexander EK, Bible KC, et al. 2015 American Thyroid Association Management Guidelines for Adult Patients with Thyroid Nodules and Differentiated Thyroid Cancer: The American Thyroid Association Guidelines Task Force on Thyroid Nodules and Differentiated Thyroid Cancer. Thyroid. 2016;26(1): 1–133. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Guth S, Theune U, Aberle J, Galach A, Bamberger CM. Very high prevalence of thyroid nodules detected by high frequency (13 MHz) ultrasound examination. Eur J Clin Invest. 2009. August;39(8):699–706. [DOI] [PubMed] [Google Scholar]
- 3.Gharib H, Papini E, Valcavi R, et al. American Association of Clinical Endocrinologists and Associazione Medici Endocrinologi medical guidelines for clinical practice for the diagnosis and management of thyroid nodules-2016 update. Endocr Pract. 2016;22(5):622–639. [DOI] [PubMed] [Google Scholar]
- 4.Cibas ES, Ali SZ. The Bethesda System for Reporting Thyroid Cytopathology. Thyroid. 2009; 19(11):1159–65. [DOI] [PubMed] [Google Scholar]
- 5.Bongiovanni M, Crippa S, Baloch Z, et al. Comparison of 5-tiered and 6-tiered diagnostic systems for the reporting of thyroid cytopathology: a multi-institutional study. Cancer Cytopathol. 2012;120(2):117–25. [DOI] [PubMed] [Google Scholar]
- 6.Bongiovanni M, Spitale A, Faquin WC, Mazzucchelli L, Baloch ZW. The Bethesda System for Reporting Thyroid Cytopathology: a meta-analysis. Acta Cytol. 2012;56(4):333–9. [DOI] [PubMed] [Google Scholar]
- 7.Nayar R, Ivanovic M. The indeterminate thyroid fine-needle aspiration: experience from an academic center using terminology similar to that proposed in the 2007 National Cancer Institute Thyroid Fine Needle Aspiration State of the Science Conference. Cancer. 2009; 117(3):195–202. [DOI] [PubMed] [Google Scholar]
- 8.Renshaw AA. Should “atypical follicular cells” in thyroid fine-needle aspirates be subclassified? Cancer Cytopathol. 2010; 118(4):186–9. [DOI] [PubMed] [Google Scholar]
- 9.Theoharis CG, Schofield KM, Hammers L, Udelsman R, Chhieng DC. The Bethesda thyroid fine-needle aspiration classification system: year 1 at an academic institution. Thyroid. 2009;19(11):1215–23. [DOI] [PubMed] [Google Scholar]
- 10.Wu HH, Rose C, Elsheikh TM. The Bethesda system for reporting thyroid cytopathology: An experience of 1,382 cases in a community practice setting with the implication for risk of neoplasm and risk of malignancy. Diagn Cytopathol. 2012;40(5):399–403. [DOI] [PubMed] [Google Scholar]
- 11.Nishino M, Nikiforova MN. Update on Molecular Testing for Cytologically Indeterminate Thyroid Nodules. Arch Pathol Lab Med. 2018;142(4):446–57. [DOI] [PubMed] [Google Scholar]
- 12.Nikiforov YE, Steward DL, Robinson-Smith TM, et al. Molecular testing for mutations in improving the fine-needle aspiration diagnosis of thyroid nodules. J Clin Endocrinol Metab. 2009. June;94(6):2092–8. [DOI] [PubMed] [Google Scholar]
- 13.Nikiforova MN, Wald AI, Roy S, Durso MB, Nikiforov YE. Targeted next-generation sequencing panel (ThyroSeq) for detection of mutations in thyroid cancer. J Clin Endocrinol Metab. 2013;98(11):E1852–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Nikiforov YE, Carty SE, Chiosea SI, et al. Impact of the multigene ThyroSeq next-generation sequencing assay on cancer diagnosis in thyroid nodules with atypia of undetermined significance/follicular lesion of undetermined significance cytology. Thyroid. 2015;25:1217–1223. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Kini SR, Miller JM, Hamburger JI, Smith-Purslow MJ. Cytopathology of follicular lesions of the thyroid gland.Diagn Cytopathol. 1985. Apr-Jun;1(2):123–32. [DOI] [PubMed] [Google Scholar]
- 16.Nikiforov YE, Seethala RR, Tallini G, et al. Nomenclature revision for encapsulated follicular variant of papillary thyroid carcinoma: a paradigm shift to reduce overtreatment of indolent tumors. JAMA Oncol. 2016. August 1;2(8):1023–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Shrestha RT, Evasovich MR, Amin K, et al. Correlation Between Histological Diagnosis and Mutational Panel Testing of Thyroid Nodules: A Two-Year Institutional Experience. Thyroid. 2016;26(8):1068–76. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Roh MH. Triage of cytologic direct smears for ancillary studies: a case-based illustration and review. Arch Pathol Lab Med. 2013; 137(9):1185–90. [DOI] [PubMed] [Google Scholar]
- 19.Wu HH, Inman A, Cramer HM. Subclassification of “atypia of undetermined significance” in thyroid fine-needle aspirates. Diagn Cytopathol. 2014. January;42(1):23–9. [DOI] [PubMed] [Google Scholar]
- 20.Shrestha RT, Hennessey JV. Cytologic subclassification of atypia of undetermined significance may predict thyroid nodules more likely to be malignant at surgery. Diagn Cytopathol. 2016;44(6):492–8. [DOI] [PubMed] [Google Scholar]
- 21.Johnson DN, Cavallo AB, Uraizee I, Tanager K, Lastra RR, Antic T, Cipriani NA. A Proposal for Separation of Nuclear Atypia and Architectural Atypia in Bethesda Category III (AUS/FLUS) Based on Differing Rates of Thyroid Malignancy. Am J Clin Pathol January 2019;151:86–94. [DOI] [PubMed] [Google Scholar]
- 22.Kato MA, Buitrago D, Moo TA 3rd, et al. , Zarnegar R Predictive Value of Cytologic Atypia in Indeterminate Thyroid Fine-Needle Aspirate Biopsies. Ann Surg Oncol. 2011. October; 18(10):2893–2898). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Gan TR, Nga ME, Lum JH, Wong WM, Tan WB. Parameswaran R, Ngiam KY.Thyroid cytology-nuclear versus architectural atypia within the “Atypia of undetermined significance/follicular lesion of undetermined significance” Bethesda category have significantly different rates of malignancy. Cancer Cytopathol. 2017. April;125(4):245–256. [DOI] [PubMed] [Google Scholar]
- 24.Baca SC, Wong KS, Strickland KC, et al. Qualifiers of atypia in the cytologic diagnosis of thyroid nodules are associated with different Afirma gene expression classifier results and clinical outcomes. Cancer Cytopathol. 2017; 125(5):313–22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Bellevicine C, Sgariglia R, Migliatico I, et al. Different qualifiers of AUS/FLUS thyroid FNA have distinct BRAF, RAS, RET/PTC, and PAX8/PPARg alterations. Cancer Cytopathol. 2018;126(5):317–25. [DOI] [PubMed] [Google Scholar]
- 26.Nikiforov YE, Nikiforova MN. Molecular genetics and diagnosis of thyroid cancer. Nat Rev Endocrinol. 2011. August 30;7(10):569–80. doi: 10.1038/nrendo.2011.142. Review. [DOI] [PubMed] [Google Scholar]
- 27.Rossi ED, Bizzarro T, Martini M, Capodimonti S, Fadda G, Larocca LM, Schmitt F. Morphological parameters able to predict BRAF(V600E) -mutated malignancies on thyroid fine-needle aspiration cytology: Our institutional experience. Cancer Cytopathol. 2014; 122(12):883–91. [DOI] [PubMed] [Google Scholar]
- 28.Li C, Lee KC, Schneider EB, Zeiger MA. BRAF V600E mutation and its association with clinicopathological features of papillary thyroid cancer: a meta-analysis. JClin Endocrinol Metab. 2012. December;97(12):4559–70. [DOI] [PMC free article] [PubMed] [Google Scholar]