Clinical-pathological issues in thyroid pathology: study on the routine application of NIFTP diagnostic criteria

In 2017, the WHO classification of tumours of the endocrine organs established the criteria for a NIFTP diagnosis. The present paper considers some aspects that are still debated or unresolved: the real incidence and clinical meaning of multifocal/multinodular lesions, the biological behaviour of micro-NIFTP, the sprinkling phenomenon and the corresponding modifications to the FNA reporting systems based on changes to the ROM. Moreover, the paper suggests possible scenarios for the clinical-pathological management of this entity. From the initial 1470 cases, a group of 68 NIFTPs was recruited in a 9 year-long period. The average age at diagnosis was 55 years. The average diameter of the lesion was 1.7 cm (0.1 cm–10 cm). In 41 cases (60.1%), the lesion was inserted in the context of a multinodular background. In 12 cases, the diagnosis was incidental and the pre- operative FNA was performed on a different target. In 10 out of 68 cases, there was a multifocal NIFTP; in 14.7% of patients, PTC-like nuclear features showed sprinkling phenomenon. The cytological revision allocated 21 cases (49%) to the SIAPEC TIR3 indeterminate class and a nuclear score 2 or 3 were identified in 25 smears. Multifocality is part of the spectrum of NIFTPs, that can arise in a multinodular background with variable sizes from microscopic lesions to very large ones. Cytopathological criteria such as an evaluation of the nuclear score may help the pathologists in promoting a NIFTP diagnosis in the preoperative setting.

• Presence of capsule, thin or thick, or clear demarcation of the lesion with respect to the surrounding parenchyma and absence of vascular or capsular invasion (Fig. 1A); • Follicular growth pattern, with absence of papillae, absence of psammomatous bodies, solid/trabecular/insular growth pattern <30%, absence of necrosis and low mitotic index (less than 3 mitoses per 10 high power fields); • PTC-like nuclear characteristics, with nuclear score (NS) of 2 or 3 (Fig. 1B). The nuclear score included the evaluation of the size, shape and clarification of the tumour nuclei, as previously described 1 clinical features. For the final clinical-pathological characterisation of the NIFTP series, the following parameters were recorded: sex of the patient; age at diagnosis; maximum tumor diameter; presence/absence of lymph node metastasis at presentation; mean, median and range of follow-up in years; uninodular/multinodular background (goiter, chronic lymphocytic thyroiditis, concurrent thyroid tumours); incidental tumor; single/ multifocal lesion; NIFTP showing "sprinkling" phenomenon. NIFTP was considered "multifocal" if two or more lesion foci were present in the same thyroid lobe or in both lobes. In the case of multifocal lesions, the maximum diameter was evaluated taking into account the lesion with the largest diameter. NIFTP was considered incidental if the preoperative fine needle aspiration (FNA) was performed on a different target.
cytopathology. The original cytological preparations that referred to the preoperative FNAs were retrieved.
Traditional Pap-stained smears were diagnosed according to the 5-tiered 2014 Italian SIAPEC-IAP classification system, which is comparable to UK Royal College of Pathology system (Fig. 1C) [17][18][19][20] . To confirm the representativity of the series, the incidence of the different diagnostic class was compared to a standard, expected, distribution taken from the study of Bizzarro et al. 21 and to a target distribution of the international review by Bongiovanni et al. 16 . The cytological NIFTP NS was also calculated in blind from the histological results.
Statistical analyses. Statistical

Results
From the EFVPTC group, 7 cases were excluded due to the unavailability of the BRAF status. Of the remaining 77 cases: • 17 cases were reclassified as invasive EFVPTC due to capsular invasion being clearly present; • 10 cases were reclassified as well-differentiated tumours of uncertain malignant potential (WDT-UMP, WHO 2017) because capsular /vascular invasion was doubtful; • 4 cases showed focal papillae 22 ; Overall, 46 out of 77 cases (59.7%) were reclassified as NIFTP due to satisfying all the required diagnostic criteria. Therefore, the study consisted of a group of 68 NIFTPs in a 9 year-long period (50 female patients −74%). The average age at diagnosis was 55 years (range 21-81 years). The average diameter of the NIFTP at diagnosis was 1.7 cm (range between 0.1 cm and 10 cm, standard deviation 1.6, median 1.3 cm). In 28 out 68 cases (41%), the diameter was <1 cm; while in 4 patients >4 cm. (Fig. 1D). The FNA diagnostic class was available in 55 out of 68 cases (80.8%). Table 1 compares the distribution of the most relevant clinical-pathological variables in the general population having a thyroid nodule [22][23][24] , in NIFTP Nikiforov's study [22][23][24] and in our series. The average age of the patients included in our series was approximately 10 years greater than that of the Nikiforov series, whilst www.nature.com/scientificreports www.nature.com/scientificreports/ 1-96 months). In this period no patient encountered recurrence. In 41 cases (60.1%), the lesion was inserted in the context of a multinodular background (mainly goiter, Table 2). In 5 cases, NIFTP was associated with a cPTC (3 of them were micro PTC) and, in 1 case, NIFTP was associated with a follicular tumour of uncertain malignant potential (ft-UMP). The FNA report was available in 32 of them; in 12 cases the diagnosis of NIFTP was incidental and the pre-operative FNA was performed on a different target ( Table 2). The difference between the diameter of the 43 cases with FNA performed on target on the NIFTP (mean 1.93 cm) and the NIFTP diameter in these 12 cases (mean 0.69 cm) was investigated through a linear regression model and a negative association was found between class membership and thyroid nodule diameters. A statistically significant relationship remained when controlling for age and gender (β = −0.85, p-value = 0.039). In 10 out of 68 cases (14.7%), there was a multifocal NIFTP; 1 of them was unilateral (left thyroid lobe) whilst in 2 cases pathologists highlighted the simultaneous involvement of a thyroid lobe and isthmus (Table 2). At the histological re-examination, in 10 out of 68 cases (14.7%), PTC-like nuclear features showed "sprinkling" phenomenon ( Table 2, Fig. 2). Excluding 12 FNAs that were out of target, the cytological revision allocated 21 cases to the cytological diagnostic category TIR3 (indeterminate, 49%); 14 cases (32%) were TIR4 (suspicious for malignancy); 6 cases (14%) were in the cytological diagnostic category TIR2 (benign); 2 cases (5%) were included in the cytological diagnostic category TIR5 (malignant) [17][18][19][20] . The distribution of our 43 cases reflected the pivotal studies of Bizzarro et al. and Bongiovanni et al. without a statistically significant difference (p-value equal to 0.749 and 0.567 respectively) 16,21 (Table 3).
The cytological NS was calculated in 33 out of the 43 cases in which the FNA was performed appropriately on target because in 10 cases the morphological re-evaluation of the original slides was not available (Table 4, Fig. 1E).
Of the same 33 cases, the histological NS was calculated on FFPE sections: By comparing the number of cases with a different NS, evaluated on cytological smears and histological sections, a statistically significant difference was found when the exact Fisher test was performed (p-value 0.002).

Discussion
Despite the indolent nature of the non-invasive EFVPTC, the word "carcinoma" included in its name has nevertheless led clinicians to perform an aggressive treatment (total thyroidectomy and radioiodine) for many years 2 . In 2016, an international group of pathologists proposed that the term non-invasive EFVPTC should be replaced with NIFTP, outlining the morphological inclusion criteria necessary to make this diagnosis 2 . The new terminology has been accepted and introduced in the 2017 WHO Classification of tumours of the endocrine organs 1 . In our study, we analysed a retrospective cohort of patients with a previously established diagnosis of EFVPTC and patients recruited after 2016 with a diagnosis of NIFTP. A team of thyroid-committed pathologists performed the morphological revision of the histological series to obtain a final homogeneous group of 68 cases. This was followed by the correlation with the preoperative cytological SIAPEC-IAP diagnostic class and the pathological re-evaluation of the original FNAs whenever available.   25 . Due to the clearcut    Table 4. Comparison between nuclear score evaluated on cytological smear and nuclear score evaluated on histological fomalin-fixed paraffin-embedded (FFPE) sections. Abbreviations: NS = nuclear score. evidence of capsular invasion, the former carcinoma diagnosis was confirmed in a significant rate of patients (22%), while 10 out of 77 EFVPTCs were reclassified as WDT-UMP because capsular or vascular invasion was doubtful. In those situations, although the application of the morphological criteria was rigorous, doubts persisted especially with respect to the architectural criterion "complete capsule /distinct separation of the lesion from the surrounding parenchyma" because of initial/incomplete penetration phenomena (mushrooms). The board reconfirmed that a correct sampling of the lesion is essential for a complete morphological evaluation along with a scrupulous examination of the tumour-capsule interface/surrounding-tissue interface, such as the protocol proposed by Seethala et al. 5 . The diagnosis of NIFTP in our study occurred in a prevalent female population (74%) with an average age of 55 years, 10 years older than the multicentric series of Nikiforov's that enrolled cases with diameter larger than 1 cm and were generally unifocal 2 . Effectively, these characteristics are related with an earlier age of diagnosis; moreover, the reference study on NIFTP started from second-level international institutions that selectively enroll neoplastic patients for the surgical treatment with higher aggressiveness or higher tumour-stage. On the other hand, our working group planned a monocentric study from a first-level general hospital that is primarily focused on benign thyroid pathology in patients with a more advanced age. In particular, a direct consequence was the unexpected high rate of not-solitary NIFTPs (60.1%). This percentage differs from other previously reported data that associated a NIFTP diagnosis to a "single nodule" clinical presentation; however more recent papers revealed more similar percentages with goiter or chronic pseudonodular lymphocitic thyroiditis as concurrent diseases 8 . We also recalled that a significant proportion of them were incidental (29% of multinodular cases) and a significant fraction subcentimetric (41%). The treatment of incidental thyroid tumours is still debated: some authors suggest a conservative surgical approach while Maturo et al. suggest total thyroidectomy because these tumours are commonly multifocal and/or bilateral and show occult lymph node metastasis 26,27 . In their study, Gartland and Lubitz reported that lobectomy and total thyroidectomy yield comparable oncologic outcomes for PTC measuring 1 to 4 cm, including cases of NIFTP [28][29][30] . Concerning the behaviour of the so-called micro-NIFTP (with a diameter of less than 1 cm), the literature showed a morphology and an indolent behavior very similar to that of the larger NIFTPs 3,31-33 . Micro-NIFTPs may occur as incidental findings in thyroidectomies performed for struma or other pathological conditions. On the contrary, for very-large NIFTPs (>4 cm) it was questioned whether the only surgical treatment without completion with radioiodine and TSH suppression therapy was a possible option, since the American Thyroid association guidelines suggests the combination of radioiodine therapy and surgical treatment 34 . Xu et al. reported the absence of disease recurrence in 37 patients with NIFTP over 4 cm in size that were not treated with radioiodine, and this result was confirmed by a study by Rosario et al. 3,33,35 . Therefore, NIFTPs of diameter greater than 4 cm are also characterised by having an indolent course after surgical treatment alone, which can therefore be the treatment of choice without the need to add a therapy with radioiodine and/or suppression of TSH. In our experience, we didn't observe any recurrence or disease progression in both pre-2016 radioiodine-treated pT2/pT3 EFVPTCs and post-2016 radioiodine-free NIFTPs.
Comparing cases that had preoperative FNA performed on target lesions Vs incidental NIFTPs, diameters of these two subgroups showed a statistically significant difference (p-value 0.039). In a previous study of our group that investigated the problem of incidental diagnoses in a PTCs series, the frequency of incidentalomas accounted for 45.2% 25 . Accidental diagnoses were more associated with size and a multinodular clinical presentation; both small (difficult to detect) and large (so-called prevailing) nodules were possible pitfalls 25 . In wide lesions (>4 cm) FNA-sampling only occurs in a limited area of the tumour, whilst the classical NIFTP diagnostic nuclear alterations may be present only in partial areas (sprinkling). In the pre-NIFTP era, Vanzati et al. have described cases of Follicular Variant of PTC (FVPTC) that, although at a lower magnification might appear as follicular adenomas (FA) or hyperplastic nodules, at high magnification presented microfollicles with PTC nuclear characteristics 10 . These modifications could be correlated with a genetic multifocality that is developing from a hyperplastic nodule or from a FA 10 . Compared to the multifocality parameter, 14.7% of our NIFTPs were plurifocal, as recently described 6,8 ; 70% of these were bilateral. Multifocal NIFTPs are similar in appearance to multifocal classical PTC but, whilst some multifocal PTCs may derive from an intrathyroid lymphatic diffusion, this could not be true for NIFTP cases given that, by definition, they cannot exhibit lymphatic invasion (Fig. 3) 9 . Thompson et al. showed that this did not result in disease recurrence in patients treated and undergoing follow-up 6 . Schmitt recently underlined that multifocality and bilaterality are part of the spectrum of NIFTP, so active surveillance of the contralateral lobe should be considered when a conservative surgical approach is chosen 8 . Data regarding the clonal relationship between different synchronous intrathyroid NIFTP foci is poor and this aspect could be studied in future genetic projects. The decrease in terms of prevalence of PTC undoubtedly affects the ROM associated with diagnostic cytological classes of FNAs 3,7,11-16 . In our series, the majority of NIFTPs were classified at FNA in the indeterminate (TIR3) category (50%) and the second most frequent diagnostic class was suspicious (TIR4, 32%). The remaining cases were distributed in the benign and malignant diagnostic categories, with a slight prevalence of the benign category compared to malignancy (14% Vs 5%).The more recent updating of the italian SIAPEC system for reporting thyroid pathology, encourage to use the sub-category Tir 3b with reference to NS2/3 when referring to the possibility of a NIFTP on cytological material (the equivalent category could be THY3F for the British classification) 22 . This would facilitate appropriate programming of surgical therapy. It should be noted that the percentages of NIFTP distribution within the indeterminate cytological diagnostic categories reported in the various literature studies are very variable: from 13% to 57% for the atypical cells of undetermined significance -AUS-/follicular lesions of undetermined significance -FLUS-category of the Bethesda System for Reporting Thyroid Cytopathology, from 13% to 82% for the follicular neoplasms (SFN/FN) category and from 13% to 35% for the suspicious of malignancy 7,[12][13][14]21,36,37 . This variability is attributable to several factors, such as the lack of safe cytological criteria with reliability and high inter/intra-observer reproducibility, the lack of a defined diagnostic class in which to insert the NIFTP and the number of NIFTPs that are diagnosed at an institution. In fact, the comparison of our case study with that of another Italian centre that took into account a similar number of cases (43 versus 37), and with the review of Bongiovanni et al., did not show statistically significant differences conclusions The NIFTP era is changing the approach to thyroid histopathology; physicians should envision a NIFTP diagnosis not only for solitary nodules, having important consequences both in the ecographic triage and in the pathological examination. Multifocality and bilaterality are part of the spectrum of NIFTPs, that can arise in a multinodular background with variable sizes from microscopic lesions to very large ones. In this sense, a close monitoring of the contralateral lobe is justified in patients that are not submitted to total thyroidectomy. Moreover, the heterogeneity of PTC-like nuclear features is a significant factor that may affect histopathology. The possible coexistence of two lesions and the so-called sprinkling phenomenon should be considered carefully. Finally, although a NIFTP diagnosis may be incidental (especially in a multinodular goiter presentation, where FNA is performed on a different target), cytopathological criteria such as the NS evaluation may help the pathologists in including NIFTP in the differential diagnosis of indeterminate cases in the preoperative setting and to more appropriately select the correct treatment.