Multi-modal Ultrasonic Diagnosis and Prognosis for Primary Thyroid Lymphoma Invasiveness

Objective Sixty-nine patients with primary thyroid lymphoma (PTL) were evaluated for diagnostic value and prognosis for PTL invasiveness. Methods We retrospectively (2008–2019) analyzed multi-modal ultrasonic images and clinical characteristics from pathologically conrmed PTL patients. These patients were divided into aggressive PTL(n=46) and indolent PTL(n=23). Results Age(>70 years old) and elevated LDH (lactase dehydrogenase) were statistically different clinical features between aggressive and indolent PTL. From ultrasonic images, 34 cases were nodular, 11 diffused, and 24 mixed. Mixed types displayed high invasiveness (45.7%) while diffuse types displayed higher inertness (39.1%), and differences were statistically signicant (P = 0.000). Elastography, invaded thyroid capsule and increased chaotic vascularity also showed signicant differences between aggressive and indolent PTL. We observed statistical difference in OS(overall survival rates) between aggressive and indolent PTL(p=0.032). Single factor Kaplan-Meier (K-M) analysis showed that: age > 70 years old, aggressive pathology, Ki67>30%, elastography scored >3 were positively correlated with the risk of poor prognosis of PTL (P < 0.05). Conclusions Multi-modal ultrasound provides accurate ultrasonographic information, e.g., ultrasound patterns, elastography, invaded thyroid capsules, and hypervascularity, which facilitates PTL invasiveness diagnostics for improved clinical treatment. In addition, PTL patients with age > 70 years old, aggressive pathology, Ki67>30%, elastography scored >3 are more likely to have poor prognosis.


Introduction
Primary thyroid lymphoma (PTL) is rarely observed, accounting for 2-5% of primary thyroid malignancies [1,2]and 2% of all extranodal lymphomas [3]. PTL is usually B-cell-derived Non-Hodgkin lymphoma, the most common type being diffuse large B cell lymphoma (DLBCL) and followed by mucosa-associated lymphoid tissue lymphoma (MALT). Currently, a PTL diagnosis mainly depends on pathology, however, its preoperative diagnosis accuracy is not high. PTL is often misdiagnosed as thyroid carcinoma, Hashimoto's thyroiditis (HT), nodular goiter, and other diseases. Multi-modality ultrasonography including elastosonography, blood ow signal, and contrast-enhanced ultrasonography (CEUS) are useful non-invasive techniques widely used for liver, breast, and thyroid diagnostics [4][5][6].
Multi-modal ultrasonic and clinical features from 69 patients were pathologically con rmed as PTL, and retrospectively analyzed in this study. We sought to improve the diagnostic accuracy of PTL, explore the diagnostic value of ultrasound classi cation for aggressive primary lymphomas, and perform prognostic analysis.
Materials And Methods guidelines [11]. During the process, lesion location was continuously and dynamically observed from entering the lesion to completely disappearing in almost 2 min.

Clinical features
Patient pathological types comprised non-Hodgkin's lymphomas, with 46 aggressive PTL and 23 indolent cases (Table 1). Of these, aggressive lymphomas included ve Burkitt lymphoma cases (Figs. 1), four transformation type cases (also called mixed types; DLBCL with MALT), 35 DLBCL cases (Fig. 2), and two follicular lymphoma (FL) grade IIIa (FL IIIa) cases. Indolent lymphoma comprised 23 MALT cases (Fig. 3). The age of aggressive lymphoma was older in patients, mostly over 70 years old, and was statistically different to indolent PTL. Aggressive lymphoma was usually de ned as ki67 > 30% [12] and was consistent with the pathological type (P = 0.000). Of these, the pathological types of ve cases were complicated by papillary thyroid carcinoma (PTC), four cases with MALT, and one case with DLBCL.
Forty-two cases (61%) had chronic thyroiditis (Hashimoto's disease). Approximately 26% of patients with primary lymphoma had hypothyroidism, and 91% had aggressive PTL. Patients with aggressive lymphoma had higher lactate dehydrogenase (LDH) levels, which were statistically higher than patients with indolent lymphoma.

Multi-modal ultrasound features
Of the 69 pathological cases, 22 underwent ultrasound-guided needle biopsy and 47 cases had pathological tissue from surgical procedures. Ultrasonic characteristics are shown ( Table 2). We observed that ultrasonic features of nodular thyroid lymphoma, with different pathological types, included signi cant internal hypoecho, internal cord-like or grid-like hyperecho, enhanced posterior echo, a large tumor diameter of > 2 cm, a non-imbalanced aspect ratio, less internal calci cation, and blood ow signals in the mass. Ultrasonography had a higher positive predictive value for aggressive lymphoma. PTL was divided into three ultrasonic features; diffuse, nodular, and mixed types, based on Ota et al. [13]. Of these, 19 cases were diffuse, 41 nodular, and 14 mixed types. Based on our data, the ultrasound patterns of aggressive and indolent thyroid lymphomas were statistically different (P < 0.05). For mixed ultrasound classi cations, the proportion of aggressive thyroid lymphomas (45.7%) was signi cantly higher than indolent lymphomas (13%, P = 0.000), while for diffuse ultrasound classi cations, the proportion of aggressive thyroid lymphomas (4.3%) was signi cantly reduced when compared with indolent lymphomas (39.1%, P = 0.000). The thyroid capsule invasion of invasive lymphomas was more likely to be detected (P = 0.007). Blood ow signals of nodules in aggressive PTL were higher than for indolent PTL. From nodular ultrasound classi cations, most lesions in aggressive lymphomas were hard, while indolent lymphomas were softer (P = 0.000). Lesions in CEUS were diffuse, non-uniform, mildly enhanced, and some echo-free space was seen locally (Fig. 3). Unless otherwise indicated, data are numbers of patients, with counted percentage in parentheses. †Data are as mean ± standard deviation, with ranges in parentheses.
Single factor Kaplan-Meier (K-M) analysis showed that: age > 70 years old, aggressive pathology, Ki67 > 30%, elastography scored > 3 were positively correlated with the risk of poor prognosis of PTL (P < 0.05).These results are shown in Table 3. However, there was no evidence to show that the gender, elevated LDH, hashimoto's thyroiditis, diameter, abnormal lymph nodes, invaded thyroid capsule, asymmetric enlargement and blood ow were signi cantly associated with the prognosis of PTL (P > 0.05) using single factor K-M survival analysis. The characteristics (P < 0.05) were used to construct a multivariable cox regression equation and the results showed that age (Hazard Ratio(HR) = 15.62, 95% Con dence Interval (CI) 1.79-136.64, P = 0.013) was independent risk factor for PTL. These results are also shown in Table 3.

Discussion
Almost all PTLs are B-cell lymphomas which express the B-cell markers, CD20, CD79a, and Pax-5, whereas T-cell lymphomas are rare [14]. Our cases were all non-Hodgkin's lymphoma, and the proportion of female patients was relatively high, with an average age of 62 ± 11 years. The age of aggressive PTL was older when compared with indolent PTL, and most had elevated LDH levels which was consistent with previous reports [1,15]. Therefore, when examining older patients at initial consultation for Hashimoto's disease, clinicians must be aware of a high risk for PTL in these patients. DLBCL and MALT tumors were the most common PTL pathological types. In this study, the proportion of DLBCLs was the highest (51%), followed by MALT (33%). MALT-type lymphomas were mostly localized lesions, with low malignancy and a good prognosis, while DLBCLs were prone to metastasis and highly malignant. Burkitt lymphoma and follicular lymphoma are rare in PTL; their histological appearance and immunophenotypes are basically consistent with intra-lymph node lesion characteristics [12]. DLBCL combined with MALT indicated a transformed lymphoma, which was more aggressive than DLBCL alone, suggesting MALT lymphoma could be transformed into higher-grade, more aggressive lymphomas [16].
The mainstay treatment is combined modality therapy (CMT) for aggressive lymphoma and radiotherapy(RT) for localized indolent lymphoma. It was reported that CMT improved the prognosis of patients with aggressive lymphoma, while CMT for indolent lymphoma remains uncertain [3]. Multi-modal ultrasound follow-up may be helpful for early differentiated aggressive PTL from indolent PTL among patients with Hashimoto's disease.
PTL exhibits characteristic sonographic and doppler ow features which show central blood ow patterns highly suggestive of PTL rather than nodular goiter [17]. However, some sonographic features, such as asymmetric goiter with marked hypoechogenicity, internal cord-like or grid-like hyperechoic, enhanced posterior echo, large size, absent calci cation and hypervascularity, and aspect ratio is not misaligned, etc., are different from most solid malignant lesions [18,19]. While PTL is rarely observed, my study had focused on multi-modal ultrasound. Cases in this study were divided into nodular, diffuse, and mixed types based on ultrasound image characteristics from previous reports [20]. The mixed type was more prone to aggressive lymphoma, and was mainly based on lesion distribution and morphology, suggesting different PTL types may be manifested as different stages of lymphoma growth and development [17,21]. A previous study [22] reported that CEUS for PTL indicated a diffuse, non-uniform lling type, just like the snow ake enhancement observed in our study. Yang et al. [23] reported that CEUS was an e cient diagnostic tool for the differential diagnosis of PTL and nodular Hashimoto's thyroiditis for patients with diffuse HT. Previous reports have focused on nodular elastography for PTL. Wang et al.
reported that along with several other parameters, a central blood ow pattern was highly suggestive of a PTL diagnosis rather than nodular goiter [17]. Nodular blood ow signals in aggressive PTL exhibited higher grades (usually 2-3) than indolent PTL (generally 0-2) in our study.
Previous studies showed that for thyroid lymphoma, core needle biopsy (CNB) sensitivity and accuracy were higher than ne-needle aspiration(FNA) for both preoperative and postoperative patients, and that CNB reduced unnecessary diagnostic surgery [24,25]. In our study, four patients underwent FNA, suggesting that malignant tumors and lymphomas were to be examined, and then CNB was diagnosed, suggesting that conventional thyroid FNA may be limited in the diagnosis of PTL due to the limited number of cells obtained and the limitation of immunohistochemical technology. In addition, multi-modal ultrasound provided more information on PTL lesions for puncture, therefore CNB guided by multi-modal ultrasound is more advantageous for diagnosing PTL when compared with FNA or surgery.
PTL combined with thyroid cancer is very rare [26,27]. In this study, we observed ve PTL cases with PTC, maximum age 62, and four cases are MALT lymphoma with HT and diffuse type of ultrasound pattern.
Interestingly, the prognosis was not worsened by combined PTC and PTL, but was more likely affected by the more poorly staged tumor. Thus, for patients with concomitant PTC and PTL, treatments were conducted which prioritized the tumor with the worse prognosis and/or the worse stage at diagnosis[28]. Graceffa et al. [29] expounded the frequent coexistence of PTC and HT. Similarly, the probability of PTL in HT populations was approximately 50 times higher than the general population [1,9]. This situation highlights the importance of monitoring HT patients, particularly if they have long-lasting disease with nodules. Patients with coexisting PTC and PTL require a multidisciplinary approach for diagnostic workup and optimal management.
PTL has a good prognosis, which is closely related to the pathological type and clinical stage, especially tumor size and extent of cancer invasion. MALT is an indolent lymphoma with a 5-year survival rate of 62-96%, while DLBCL is an aggressive lymphoma with a 5-year survival rate of 45-75% [30]. Cha et al. [31]reported that the 5-year OS and PFS rate of aggressive PTL was 75.6% and 73.9%, respectively. Watanabe et al. [32]found that ve-year overall survival and event-free survival for elderly patients with thyroid DLBCL were 87% (95% con dence interval [95% CI], 64-96%) and 74% (95% CI, 50-89%), respectively. In this study, the OS of indolent PTL patients was signi cantly better than aggressive PTL patients(P < 0.05). PTL patients with age > 70 years old, aggressive pathology, Ki67 > 30%, elastography scored > 3 are more likely to have poor prognosis. Multivariable cox regression analysis showed that age was an independent risk factor for PTL. Watanabe et al. [33] in another research reported that higher age and higher erythrocyte sedimentation rate (ESR) were signi cant risk factors for OS, which is almost similar to our research.
Our study had limitations. Firstly, PTL is rare and the single-center nature of our study meant our cohort was small. Secondly, to precisely identify histopathological PTL types over a long period, only patients with a pathologically proven diagnosis were included in this study. Also, some cases were lost to followup. However, PTL is relatively rare, therefore, it is di cult to perform prospective studies and identify large numbers for analysis.
In summary, the application of multiple ultrasound examination modes is bene cial for capturing disease information and PTL ultrasound patterns. Also, the approach comprehensively evaluates tumor boundaries, sizes, shapes, internal echo, hemodynamics, spatial structures, soft textures, and pathological tissues. In addition, PTL patients with age > 70 years old, aggressive pathology, Ki67 > 30%, elastography scored > 3 are more likely to have poor prognosis. It is recommended elderly patients with HT are closely followed-up by multi-mode ultrasound to improve PTL early detection and strategically target aggressive tumors.