Although of a low incidence, thyroid lymphoma (TL) and anaplastic thyroid carcinoma (ATC) are aggressive malignancies in need of urgent management. Presentation is similar (usually a 60-70-year-old female with a rapidly enlarging neck mass over weeks to months, and with symptoms like neck pain, dysphagia, hoarseness and stridor), but prognosis and management differ (1–6). Urgent management would benefit from an evidence-based strategy excluding unnecessary delays in reaching a precise diagnosis.

TL accounts for 0.4-5% of thyroid malignancies and for less than 2-3% of all extranodal lymphomas (3, 7–9). Most TLs are non-Hodgkin B-cell lymphomas. The most common subtype (>50%) is diffuse large B-cell lymphoma (DLBCL), followed by mucosa associated lymphoid tissue lymphoma (MALTL; about 10–30%) (3, 10). T-cell and Burkitt thyroid lymphomas are extremely rare (11, 12). Subtype classification is of great importance as prognosis and therapy depend on the lymphoma histotype. MALTL tends to be less aggressive than DLBCL and T-cell lymphomas. The latter two need more intensive chemotherapy or chemoradiotherapy (10). Overall, TL responds well to non-surgical treatment with a disease-specific survival of up to 79% (1). In contrast, ATC responds much less to treatment (5). While accounting for <2% of thyroid malignancies, ATC accounts for 25% of all thyroid cancer-related deaths (5, 13). In selected ATC, where R0, or at most R1 resection with acceptable morbidity seems likely, extensive surgery needs prompt adjuvant intensive radiotherapy and/or chemotherapy (14); when surgery is not an option, primary chemoradiotherapy, palliative radiation, systemic therapy, or best supportive care is offered. Prognosis is poor, with a median survival of 5 months and a 2-year overall survival of 10% (5, 13). Regardless of the treatment, most patients have a rapidly growing, extensively locally invading tumor, causing death with uncontrolled local disease and distant metastases. One series describes 36% of their patients dying of airway obstruction, the remainder succumbing to generalized disease progression (13, 15). Nonetheless, recent molecular-based personalized treatment strategies achieve better cure rates, with the combination of dabrafenib (inhibitor of BRAF) and trametinib (inhibitor of MEK) now being FDA (U.S. Food and Drug Administration)-approved in ATC patients with the BRAFV600E mutation (5, 16).

Imaging studies in the acute setting consist of ultrasound (US) as well as cross-sectional imaging. US typically shows a heterogeneous echogenicity, a diffuse infiltration into surrounding tissues, an irregular shape and an increased vascularity. Calcifications would be more in keeping with ATC, whereas the presence of echogenic strands and enhanced posterior echoes are more indicative of TL (3, 17). Cross-sectional imaging is essential to visualize the relationship of the tumor to the major vessels, trachea and esophagus, which guides surgical decision making. Computed tomography (CT) is preferred over magnetic resonance imaging (MRI): the shorter data acquisition time in patients with often some form of respiratory distress results in more accurate images. CT may show calcification, necrosis and heterogeneous attenuation in ATC; homogeneous attenuation without calcification or necrosis suggests TL (6). On MRI ATC shows also a moderate-to-marked heterogeneous enhancing with central non enhancing areas of necrosis, and mixed signal on T1- and T2-weighted images, whereas TL tends to be homogenous, similar to the CT aspect of TL (18).

Tissue analysis using US-guided cytology and/or histology is essential for establishing an accurate diagnosis. In this context, FNA, generally performed using a 21-25 Gauge needle on a 10 cc syringe under ultrasound guidance, with up to 8 needle passes, depending on the on-site judged appropriateness of the obtained material, is often the initial diagnostic tool (5, 19, 20). The real-world reported sensitivity of FNA is 48% for TL (21) and 50-61% for ATC (22, 23); for ATC an FNA-based correct diagnosis in >60% requires immunohistochemistry (IHC) on cell-blocks from cell-rich aspirates (24). These low figures are attributable to a high rate of false negative results, related to the pathogenesis of both malignancies. TL often develops in a background of Hashimoto’s thyroiditis (HT), chronic autoimmune stimulation being considered as the main trigger for malignant transformation (7, 25–27). The proportion of TL with a known history of HT is 80%, conversely, compared to the general population, patients with HT have a 60-fold increased risk of developing TL (28, 29). Distinguishing MALTL from HT is particularly difficult on cytology; both conditions may coincide and their lymphocytic populations show a similar morphology (30, 31). ATC may show extensive necrosis (32) and it causes little surprise that FNA often misses viable tumor.

Resort to US-guided core needle biopsy (CNB), generally performed under local anesthesia and ultrasound guidance, stabbing 2-3 times using a 16-20 Gauge spring-activated 2 cm excursion needle, comes currently after (persistently) inadequate results with FNA (5, 13, 20, 33). Nonetheless, Na et al. recommended immediate CNB in patients with thyroid masses with uncommon clinical and radiological features; the recommendation was based on a small amount of retrospective cohort studies, not specific for ATC and TL (34).

CNB aside, surgical incisional biopsy (IB; also referred to as “diagnostic surgery”; making a skin incision under local or general anesthesia to take a substantial tissue sample (2-3 cm³) of the tumor, avoiding the necrotic parts) would seem the obvious choice in overcoming the FNA inadequacies. Enough tissue of preserved architecture is thus obtained to effect routine histology and ancillary studies, including IHC for lymphoma subtyping and, defining molecular markers for ATC such as Ki-67, p53, BRAF V600E mutations and eventual NTRK, ALK or RET fusions (5, 24, 25, 35–38). However, IB in patients with TL or ATC carries significant risks. The procedure is invasive, time-consuming and involves hospital admission, general anesthesia and intubation of patients often already in respiratory distress. Extubation is unsure, whereas tracheotomy upon failed extubation is hazardous or even impossible, due to the compressive thyroid mass. Moreover, wound healing of the incision is often compromised ( Figure 1 ).

Accordingly, to reduce the risks of IB and to minimize the delay to treatment, CNB seems a diagnostic tool more sensitive than FNA. Less invasive than IB, CNB allows for the same adequate IHC and molecular marker analysis. For diagnosing malignant lymphomas in the neck outside the thyroid, as well as for the more common thyroid neoplasms, CNB has yielded satisfactory results with reported sensitivities of up to 100% (35, 39, 40). In the narrower context of diagnosing ATC and TL, a thorough review of the reliability of CNB is lacking. This prompted the present systematic review and meta-analysis, to assess the value of CNB in diagnosing ATC and TL by calculating sensitivity, PPV, and the rate of diagnostic surgery.

For obtaining cells or tissues for microscopical examination in the situation of a patient presenting with a rapidly enlarging thyroid mass, FNA is the instinctive first test to be performed, but its limitations and low sensitivity have already been observed. In this context, the present meta-analysis suggests a superior diagnostic value of CNB, with a sensitivity and PPV of 94.3% and 100% for TL, 80.1% and 100% for ATC. The meta-analysis also suggests that CNB enables reducing the need for diagnostic surgery to 6.2% for patients with TL, to 17.6% for ATC and to 7.4% for both entities. In comparison, the need for IB after FNA is reportedly 34% for patients with ATC, 37.9% for TL, and 35.4% for both entities (40).

Besides more accurate typing a clear advantage of CNB over FNA is the more reliable IHC/genetic analysis. This is especially relevant since recent series using targeted therapy in BRAF V600E-mutant ATC have reported promising results, and a substantial clinical benefit (OS of 31,5% at 2 years) through dabrafenib and trametinib (37). In the absence of BRAF V600E mutation, ALK, RET and NTRK fusions, when present, can be targeted; checkpoint inhibitors have their place in case of high PD-L1 expression (5).

Risks of IB have been mentioned – hence this procedure is best reserved for the occasions where CNB is inconclusive. The main arguments of specialists who are reluctant to implement CNB are the requirement of a radiologist and imaging infrastructure on site, local anesthesia, a higher cost, more discomfort, risk for tumor seeding and a higher bleeding risk (45). However, looking critically at the literature, only self-limiting hemorrhages were reported, the incidence of post-CNB hematomas ranging between 0.02% and 2,4% (58, 60).

The present first systematic review and meta-analysis assessing the diagnostic value of CNB in patients with ATC and TL indicates the high sensitivity and PPV of the procedure, the low need of diagnostic surgery and rare complications. It thus seems sensible to infer that CNB is the most appropriate initial diagnostic tool for obtaining cells or tissue for microscopical examination. Based on the flowchart suggested by Wu et al. (56) and the results of the present meta-analysis, an updated flowchart is now suggested ( Figure 4 ). Recently in April 2022, ten months following the literature search of the current meta-analysis, an Italian prospective series compared FNA and CNB in the patient group presenting with a rapidly growing thyroid mass. The findings in the current meta-analysis were corroborated by this report on 13 TL and 33 ATC, worked-up in an experienced national tertiary referral center for thyroid cancer. In ATC, FNA was able to correctly identify histotype in only 3 patients (12.5%; PPV 17.1%) whereas CNB did so correctly in all 33 (PPV 100%). FNA correctly diagnosed TL in 7 of 13 patients (53,8%; PPV 61.5%), whereas CNB correctly diagnosed 12 of 13 TL (92%; PPV 92.3%) (20).

The obvious limitations to this systematic review and meta-analysis follow the inherent retrospective nature of the included retrospective studies dealing with two rare disease entities. Selection bias is difficult to eliminate. Ideally, studies should share the same inclusion criteria. First, however, the limited available literature necessitates allowance of studies with broader inclusion criteria than just TL or ATC, when they do offer the necessary raw data; these were then selectively extracted according to predefined inclusion criteria. Second, also of studies focusing on ATC and TL, inclusion criteria differed between the included manuscripts; most studies did not perform CNB primarily but only after inconclusive FNAs, and mostly CNB was performed in less patients than FNA; this selection bias may underestimate the diagnostic value of CNB, as evidenced in the study by Matrone et al. (20)

Also, following the inclusion criteria of many of the studies selecting only patients with TL and ATC, specificity and NPV could not be calculated for these patients, and the PPV of 100% should be interpreted with caution.

Finally, not all studies provide a detailed description of the CNB technique, nor of the experience of the radiologists and pathologists involved, which is likely to be quite divergent; these variables are thus not reflected in the findings of this meta-analysis.

For patients presenting with a rapidly growing thyroid mass, CNB has a significantly higher sensitivity for in the diagnosis of ATC (80.1%) and TL (94.3%) than FNA (61% and 48%, respectively). Not associated with increased complications, and facilitating an earlier treatment, CNB thus seems the preferred initial step to obtain tissue diagnosis and start adequate treatment in this clinical scenario.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding author.

This paper was written by members and invitees of the International Head and Neck Scientific Group (http://www.IHNSG.com). VV, NG and AF contributed to conception and design of the study. VV and NG organized the database and performed the statistical analysis. VV, NG, and AT wrote the first draft of the manuscript. VV, NG, AT, MC, PC, OC, PG, OG-L, CP, GR, AR, OR, AS, ARS, YT, RT, MW, MZ and AF wrote the final draft of the manuscript. All authors contributed to the article and approved the submitted version.

Supported by the Walter Vandeputte Head and Neck Cancer Fund (KU Leuven, Leuven, Belgium).

The authors would like to sincerely thank Annouschka Laenen, PhD, from the Leuven Biostatistics and Statistical Bioinformatics Centre, KU Leuven, Leuven, Belgium, for the help in the statistical analysis.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.