Ultrasonography‐Guided Core‐Needle Biopsy for Newly Diagnosed Large B-Cell Lymphomas: Analysis on Diagnostic Efficacy and Safety in an Italian Retrospective Study Open Access

The updated 5th Edition of the World Health Organization (WHO) classification has refined the categorization of large B-cell lymphomas (LBCLs), which are a heterogeneous collection of clinic-pathological entities, of which diffuse LBCL not otherwise specified (DLBCL, NOS), and high-grade B-cell lymphoma with MYC and BCL2 rearrangements subtypes are the most common [1]. LBCLs are very aggressive diseases and require prompt diagnosis to rapidly start adequate treatment [2]. Studies regarding the decision of which primary interventionist diagnostic tool to use for suspected LBCLs are clinically relevant and called for. For this purpose, new mini-invasive approaches based on imaging-guided methods are now available [3]. The few clear indications for performing such a procedure according to existing guidelines have raised questions regarding its efficacy and widespread adoption [4]. However, the introduction of a new generation of ultrasonographic and biopsy needle devices, which have already been proven valuable in the management of patients with lymphoma [5‒8], provides the opportunity to develop an effective combined diagnostic strategy. Thus, the lymph node core-needle biopsy (CNB) under imaging guidance in the pathologic diagnosis of LBCLs requires validation. In a large-scale retrospective survey of the French Lymphopath network [9], LBCLs were more frequently diagnosed in the CNB-cohort (42.4%) than in the surgical excision-cohort (32.3%) (p < 0.001). The reproducibility among pathologic reviewers for diagnosing LBCL cases was high and similar on CNB and surgical excision (86% and 82.2%, respectively). However, CNB resulted in a higher number of cases with no diagnosis due to insufficient or inadequate material (1.8% and 0.1% with CNB and surgical excision, respectively; p < 0.001). CNB provided a definitive diagnosis of lymphoma according to the WHO Classification in 92.3% versus 98.1% of surgical excision biopsy (p < 0.001). In the French collection [9], the information regarding the program for optimal imaging guidance, and for characteristics of biopsy needle, i.e., gauge, length, tip configurations, and sampling mechanisms was recorded in a minority of patients; the authors suggested to collect these data in a large cohort.

We reported in Annals of Hematology of 2017 [10] encouraging single-center efficacy results of a prospective trial in which 376 patients with lymph node enlargement suspected of lymphoma were randomly assigned to biopsy with either standard surgical biopsy or power Doppler ultrasonography (PDUS)-guided 16-gauge modified Menghini CNB. A total of 187 patients received the allocated surgical biopsy by surgeons experienced in lymph node resection, and a total of 189 patients received the allocated PDUS-guided CNB by members of the hematology staff with extensive expertise in lymphomas management and interventionist PDUS [5‒7]. Four (2%) patients undergoing CNB were excluded for inadequate histological samples; thus, 372 patients were finally analyzed. The sensitivity rate of lymph node malignant status was 88.7% for surgical biopsy with a false-negative rate of 10.2% (19 of 168 patients with lymph node positive for malignancy were not identified). By contrast, the sensitivity rate of lymph node malignant status was 98.8% for PDUS-guided CNB with a false-negative rate of 1.1% (2 of 174 patients with lymph node positive for malignancy were not identified). The presence of satellite enlarged reactive and/or necrotic lymph nodes impaired the success of surgical biopsy, whereas CNB was able to sample enough diagnostic pathologic tissue directly and accurately from viable tumor under PDUS guidance. However, the exclusion of 4 patients undergoing CNB due to inadequate histologic samples may introduce bias in the reported sensitivity rates, limiting the generalizability of the trial results.

Considering the limitations of our previous report [10] for a small cohort of LBCLs, we subsequently checked our databases for an adequate period of time in order to provide a large series of cases undergoing PDUS-guided CNB. The primary endpoint of the present study was to define PDUS-guided CNB diagnostic accuracy and safety for diagnosing LBCLs. As a secondary endpoint, we compared the LBCLs CNB findings with the CNB findings of lymphadenopathies with other etiology. Here we report the mature results from this study.

This is a retrospective study from the registry databases of our hospital units of tertiary centers (the Hematology Unit of the Federico II University Medical School of Naples and Pathology Unit of the Federico II University Medical School of Naples) in Italy from July 2009 to January 2022. In these Units, UG‐CNB was used as a routine mini‐invasive practice for lymphadenopathies suspected of harboring a lymphoma. The staff physicians were specifically dedicated to interventional procedures under ultrasonographic guidance, with >15 years of experience in the field of lymphoma diagnoses [5‒7, 11]. Data collection forms were sent to the data managers of the two centers requesting the following information: detailed facts regarding demographics of patients undergoing CNB, ultrasonographic and power Doppler examinations, target lesions characteristics (including size and site of lymph nodes), and type and size of the employed needle, number of passes for each procedure, use of any particular staining and/or special methods on biopsy specimens harvested (such as immunohistochemical or polymerase chain reaction [PCR] and/or fluorescence in situ hybridization [FISH] techniques), the final histological diagnoses (pathology reports) obtained by CNB, the methods of diagnostic confirmation (for instance, surgical biopsy), detailed information on post-biopsy assessments and complications due to UG-CNB.

All necessary approvals were obtained from the Ethic Committee of the Federico II University of Naples (Italy). The acquisition of a written informed consent was obtained from each patient to undergo UG-CNB. The form contained detailed information regarding the risk of the invasive procedure and the risk of the possible failure of the procedure because of the exiguity of the CNB sample.

First, all patients were scanned using a scanner (iU22; Philips Health-care, Bothell, WA, USA) equipped with tissue harmonic compound technology (SonoCT; Philips), power Doppler sonography, and 5-1 MHz (C5-1 curvilinear; Philips) and 9-3 MHz (L9-3 linear; Philips) broadband probes. They underwent a baseline ultrasonographic exploration of all superficial, anterior-superior mediastinum (clavicular, supra-aortic, and prevascular regions), and abdominal and pelvic lymph node areas and any abnormal (for size [long axis ≥2 cm], round shape, hilus absent, and/or hypoechoic parenchyma) lymph node was examined with power Doppler in accordance with methods already described [5‒7, 10, 11].When more than one lymph adenomegaly was present, the node selected to be biopsied presented characteristics of hypervascularization, i.e., intranodal arterial vessels with high-resistive index value (>0.6). Second, CNBs were performed by M.P., and E.V. (with more than 15 years of experience with interventional ultrasonographic procedures) [5‒7, 10, 11]. A 16-gauge diameter modified Menghini needle of 150 mm in length with automatic aspiration (Biomol HS-Hospital; Rome, Italy) was used to carried out UG-CNB with aseptic technique and cutaneous anesthesia, in accordance with methods already described [5‒7, 10, 11].

All lymph node samples obtained by UG-CNB were fixed in formalin and embedded in paraffin (FFEP), and the histologic sections were routinely stained with hematoxylin and eosin (H&E). Then, we adapted the number of antibodies (among CD2, and/or CD3, CD4, CD5, CD7, CD8, CD10, CD15, CD20, CD22, CD23, CD30, CD43, CD45RB, CD56, CD57, CD79a, CD138, and/or BCL2, BCL6, MYC, cyclin D1, SOX11, PAX5, BOB1, MUM1, Ki-67, ALK1, and/or TdT) to be used for immunohistochemical staining according to the features from H&E staining. In selected cases, B- or T-cell clonality was investigated by PCR and/or FISH. FISH and/or molecular analysis were performed on FFPE slides to detect major gene translocations using commercially available kits, such as BCL2, BCL6, MYC, Cyclin D1, and MALT1, and/or immunoglobulin gene rearrangements, and/or T-cell receptor gene rearrangements, if necessary. In situ hybridization for Epstein-Barr virus was also investigated if needed. Epithelial metastatic tumors were identified by monoclonal antibodies to cytokeratin.

Before the biopsy procedure, coagulation-screening tests were performed, and informed consent for the procedure was obtained. After the biopsy, patients were kept under observation and discharged if no signs or symptoms suggestive of a relevant complication manifested. Outpatients were closely monitored for at least 1 h after the procedure to look for CNB-related complications. All patients were encouraged to contact their physicians if they developed symptoms after leaving the hospital.

Among the patients who underwent front-line lymph node biopsy in a 12‐year period, we selected the cases with the following inclusion criteria: (i) target consisting of superficial or deep-seated de novo pathological lymphadenopathies with long axis ≥2 cm and hypervascularization at ultrasonographic power Doppler assessment, i.e., intranodal arterial vessels with high-resistive index value (>0.6) as previously described [5‒7, 11]; (ii) availability of lymph node samples obtained by PDUS-guided 16-gauge CNB (the cutting needle had a diameter of 1.6 mm with ultra-thinner tip and wall, and powered automatic suction), routinely fixed in formalin and embedded in paraffin; and (iii) core biopsy information reporting morphologic, and/or immunohistochemical features and/or FISH and/or molecular analysis (reported on the original pathologic report).

We exclude from the analysis patients who received CNB of lymph nodes with size <2 cm, with no signals of intranodal vascularization at ultrasonographic power Doppler investigation and/or distance/deep greater than 12 cm from the skin plane.

Consensus diagnosis of CNB samples was performed under the availability of an accepted diagnostic reference standard, i.e., a 3‐point work‐up including (1) consensus among pathologists on histopathological features of samples obtained by UG‐CNB, (2) diagnosis confirmation by special studies performed on CNB samples consisting of FISH and/or PCR techniques for specific lymphoma subsets, and/or (3) follow‐up surgical biopsy. All histological investigations were performed by M.M., E.V., and G.T. (with more than 15 years of experience with pathological diagnoses of specimens from interventional ultrasonographic procedures) [5‒7, 10, 11]. Tissue sections, including slides with H&E and immunohistochemical stains, and other special studies were retrospectively retrieved and then distributed to the three hematopathologists; the 5th Edition of the WHO classification was used for the final diagnoses [1]. In the Result section of the manuscript, we reported in more details how the consensus was achieved among the three pathologists, mostly regarding the cases of LBCLs according to the study design (primary endpoint: CNB diagnostic accuracy for LBCLs).

Consensus diagnosis was considered the final diagnosis of each case. Consensus diagnosis was reached in presence of at least 1 of the above reported criteria of the 3‐point work‐up. Finally, diagnostic accuracy was defined as the percentage of correct diagnoses given by each initial diagnosis of individual pathologist, compared with the consensus diagnosis.

Cost analysis for biopsy procedures was performed by adopting the perspective of the National Healthcare System. Cost calculations for PDUS-guided CNB were based on the tariffs in the Nomenclature for Outpatient Care, provided by the Italian National Healthcare System (http://www.arsancampania.it/documents/10157/01088316-4824-4c7e-8671%201418af8f3af7). The costs of surgical biopsy were calculated according to the diagnosis-related group tariffs that are currently used to fund in patient health services in Italy (http://www.eumedit/drg/tariffe_drg.asp).

Fisher’s exact test was used for the categorical variables. Continuous variables such as age, lesion size, number of slides for immunohistochemical assessment, time from the first visit to biopsy, and diagnosis were also evaluated using the Mann-Whitney U test. All tests were 2-sided, and a p < 0.05 was considered statistically significant.

From July 2009 to January 2022, among the 800 patients with lymph node CNB collected, 220 cases whose diagnoses of LBCL subtypes were first obtained by UG‐CNB were included for the primary endpoint analysis (Fig. 1). Table 1 shows the characteristics of these 220 analyzed cases with CNB diagnosis of LBCLs. The median age was 55 years (range, 22–91 years); the target lymph node lesions were in superficial sites in 70% of biopsies, and in deep-seated sites (abdomen-pelvic or antero-superior mediastinum) in the remaining biopsies; median lymph node’s long axis was 3.5 cm (range, 2–7 cm); the median waiting time for biopsy was 4 days (range, 1–10 days); the median time for US-guided CNB performance was 35 min (range: 30–50 min); the median of core-needle passes number was 2 (range 1–4); the core-needle sample median length was 35 mm (range, 15–70 mm); and median estimated volume was 250 mm³ (range, 92–430 mm³).

Table 2 shows more in detail the workflow of LBCL specimen allocation according to data collected from the initial diagnosis of individual pathologist (original pathologic report). The total number of CNB specimens collected and analyzed was 220, i.e., 220 histological blocks (1 for each patient). All of them were frontline stained for H&E. Thereafter, immunohistochemical staining was performed using a median number of ten (range, 8–15) antibodies. Overall, the total number of tissue sections used for immunohistochemical evaluations was 3,130, and the total number of tissue sections used for FISH analysis including BCL2 and/or MYC translocations was 470. The median time taken from CNB performance to reach the final histological diagnoses was 8 days (range, 7–10 days). According to the 5th Edition of the WHO classification, the specific histologic diagnoses by CNB were the following: diffuse LBCL NOS subtype (n = 196), and high-grade B-cell lymphoma with MYC and BCL2 rearrangements subtype (n = 24).

A solid basis for nonconventional histological diagnosis of LBCLs was provided by reference standard. In fact, the histological findings of malignant lymph nodes by CNB were confirmed by at least one of the 3‐point work‐up. For 16 patients, the complete surgical resection of lymphadenopathy and the subsequent conventional histological assessment confirmed the CNB diagnoses. The surgical resection of lymphadenopathies in such 16 cases likely was a patient’s decision. For 24 cases, there was confirmation by FISH techniques (including BCL2 and/or MYC translocations) for specific LBCLs subsets. Finally, except for a very minority of cases (11/220; 5%) in which consensus among pathologists occurred in 2/3 pathologists, for the remaining 169 cases (95%; the vast majority) there was a consensus made by 3/3 hematopathologists on tissue slides reviewed.

All the 220 lymph node lesions involved by LBCLs were classified in the same way by reference standard as they were by PDUS-guided CNB (Fig. 2). Consequently, the diagnostic accuracy rate of lymph node status was 100% (95% CI: 98%–100%), that is, 220 of 220 cases positive for LBCLs according to the reference standard were correctly identified by PDUS-guided 16-gauge CNB.

By using Italian values for direct costs of interventional procedures, the median cost for one LBCL diagnosis by PDUS-guided CNB was EUR 30,300 (including the US assessment of superficial and deep-seated lymph node areas, CNB procedure, and complete histological evaluations with pathological panel testing). Among the 16 LBCL diagnoses obtained by excisional biopsies, the cost for one open biopsy (including complete histological assessment with pathological panel testing) was EUR 1,050,000 for major surgery (deep-seated lymphadenopathies in mediastinum or abdomen, n = 6 cases) and EUR 320,000 for minor surgery (lymphadenopathies in superficial sites, n = 10 cases).

The flow chart and clinical characteristics of the remaining 580 cases who had lymph node positive for other malignancy subtypes (n = 510) and non-malignant findings (n = 70) who were compared with LBCL findings (secondary endpoint) are shown in Figure 1 and Table 1. We compared the CNB characteristics of the subset of LBCLs patients (n = 220) with the CNB characteristics of the patients with other malignancy subtypes (n = 510) and of the patients with non-malignant findings (n = 70), and obtained the following results at the statistical analysis: (1) the median of LBCLs lymph node long axis was significantly (p < 0.001) higher than that of other malignancy subtypes (median length: 3 cm, range 2–7 cm) and of non-malignant findings (median length: 3 cm, range 2–5 cm); (2) the median length of LBCLs CNB samples was significantly (p < 0.001) higher than that of other malignancy subtypes (median length: 30 mm, range 8–70 mm) and non-malignant findings (median length: 30 mm, range 8–70 mm; (3) the estimated volume of LBCLs CNB samples was significantly (p < 0.001) higher than that of other malignancy subtypes (median estimated volume: 150 mm³, range 92–430 mm³) and non-malignant findings (median estimated volume: 180 mm³, range 92–200 mm³).

These findings reflected that non-LBCL lesions tended to yield smaller specimens by CNB. As shown in Figure 1 and Table 1, our wide sample size of 800 lymph nodal biopsies revealed that the complex tissue architecture of non-LBCL neoplastic conditions, such as nodal peripheral T-cell non-Hodgkin lymphoma, and the stiffened tissue of Hodgkin lymphoma, which were seated in hindered regions of neck (data not shown), led to a sampling error of CNB (and thus false-negative CNB results) in 30%, and 11% of cases, respectively, suggesting that, in some instances other than LBCLs, there is a need of a large amount of lymph node tissue sampled with traditional surgical approach for correct histological assessment.

All patients, except for those biopsied in the mediastinum compartment (n = 56) who were hospitalized for a median of 2 days, underwent UG-CNB in a day hospital regimen under local anesthesia. Overall, about 30% of patients suffered from biopsy-related complications (pain, numbness or paresthesia, larger scars, lymphorrhea, hematoma, and/or wound infection).

No patients suffered from biopsy related complications of grade >2 according to the common terminology criteria for adverse events (CTCAE v.5.0; Table 3) [12]. Adverse events were always transient. Pain during the procedure was recorded as continuous but responded well to paracetamol and resolved after a median of 3 h. The hematoma within the biopsy site was painless and asymptomatic. Patients were simply monitored with ultrasonography for at least 2 h after the procedure. In none of the cases recorded the hematoma increased in size.

LBCLs, with an estimated 150,000 new cases annually worldwide, represent almost 30% of all cases of non-Hodgkin’s lymphoma. Patients typically present with progressive lymphadenopathies, extra-nodal disease, or both and require urgent histological characterization to start therapy [13]. Although lymph node imaging-guided CNB is a fast and easy mini-invasive interventional procedure [5‒8, 10, 11], the few existing guidelines focused on aggressive lymphomas diagnosis have raised questions regarding CNB efficacy and widespread adoption [3]. Recently, the Guideline by Fox et al. [14] in British Journal of Haematology (2024) describes contemporary evidence-based management of LBCLs, including new standards of baseline investigation required, or to be considered, for initial assessment of patients with LBCLs. The Grading of Recommendations Assessment, Development and Evaluation (GRADE) nomenclature was used to evaluate the levels of evidence and to assess the strength of recommendations. Regarding the front-line diagnostic work-up, the authors highlighted the importance of harvesting a good tissue sample, emphasizing the value of lymph node excisional biopsy as the gold-standard (GRADE, 1B), with the need of a tight partnership with reference hematopathology laboratory as part of a multi-disciplinary team that can facilitate surgery excisional biopsy decision making process. Importantly, they limited the use of imaging-guided CNB of lymph nodes contingent to few instances, i.e., when surgical intervention is impractical or would entail excessive risk or delay (GRADE, 2B). However, hematology institutions of international prestige from different countries rely on imaging-guided CNB to determine LBCLs diagnosis in real-life and/or in clinical trials [15‒19]. Several attempts have been made to define the best first-line interventionist diagnostic strategy to obtain pathologic lymph node tissue. There is the need to achieve accuracy of the diagnostic work-up while reducing waiting time, adverse events and costs for the performance of a biopsy, i.e., the need for a true cost-effective interventionist procedure. Table 4 shows diagnostic yields of front-line imaging-guided CNB procedures in the setting of LBCLs, by systematically reviewing the recent literature [8, 17, 18]. In our hands, a detailed imaging-guided CNB protocol was strictly followed [5‒7, 10, 11], which might have contributed to the relatively high diagnostic accuracy. Radiological-guided CNB of lymphadenopathies, with uniform program for imaging guidance (PDUS tool) and for biopsy needle (16-G modified Menghini needle), can achieve the diagnosis of LBCLs with excellent accuracy based on our wide sample size of 220 lymph nodal samples. In our study, within the standard reference, the final diagnoses were uniformly controlled by the consensus made by 2 or 3 blinded hematopathologists on reviewed CNB samples and/or confirmation by molecular biology tests including PCR and/or FISH techniques for specific LBCL subsets.

The patients had short waiting time for biopsy with a median of 4 days (range, 1–10 days), and, most important, an expedited time to reach final histological diagnosis in 8 days (range, 7–10). Outpatients were closely monitored for at least 1 h after the procedure to look for CNB-related complications. Patients were kept under observation and discharged if no signs or symptoms were suggestive of a relevant complication manifested. All patients were encouraged to contact their physicians if they developed symptoms after leaving the hospital. No patients suffered from biopsy related complications of grade >2 according to the CTCAE. Adverse events were rare. Pain and hematoma within the biopsy site resolved after a median of 3 h. Patients were simply monitored with ultrasonography for at least 2 h after the procedure. In none of the cases, the hematoma increased in size. Finally, considering the direct costs of interventionist procedures (including the complete pathological panel testing at histological evaluations), the median cost for one PDUS-guided CNB was not expensive (EUR 30,300).

Our study has limitations that must be pointed out. First, we suffer the absence of clinical follow-up data for patients who received diagnosis of lymphoma by CNB, particularly in light of findings discussed in Blood Advances by Desai et al. [20]; hence, our findings need to be validated in prospective studies with long-term follow-up data to provide a more comprehensive evaluation of CNB’s diagnostic utility and prognostic implications. Second, in our study there was a selection of patients which could compromise the reproducibility of PDUS-guided CNB procedure. In fact, in the pre-phase of collection data, we didn’t consider those patients who underwent CNB of lymph nodes with size <2 cm, no signals of intranodal vascularization at ultrasonographic power Doppler investigation and/or distance/deep greater than 12 cm from the skin plane (i.e., inaccessible or high-risk anatomical locations). In our opinion, these patients are not good candidates to CNB due to the relevant rate of failures. Third, in 5% of CNBs the consensus among pathologists for LBCL diagnoses occurred in only 2/3 pathologists. However, we considered the 2/3 rate enough to define the CNB diagnosis in such instances, underscoring the critical role of hematopathologist expertise in interpreting CNB samples. Fourth, in our series about 4% of CNB results were inadequate and/or inconclusive for lack of morphological atypia or architectural criteria at histopathology analysis (Fig. 1). Fifth, 16 patients (7%) out of 220 patients of LBCL series underwent lymph node excisional biopsy after CNB. We could not exclude that in these cases surgical biopsy was performed based on a physician decision for the suspicion of insufficient material, diagnostic uncertainty, or atypical presentations of CNB samples. However, the diagnostic failures of CNB in these 16 LBCL cases move away from that 100% accuracy rate herein reported so little, without any negative impact on the study message. Finally, our series did not include same LBCL subtypes, like for example T-cell-rich LBCL, LBCL with IRF4 rearrangement, transformation from preexisting small B-cell lymphomas, or the spectrum of gray-zone lymphomas in mediastinal locations, which are all potential pitfalls of CNB method. In our series, through uniform program for imaging guidance and characteristics of biopsy needle, PDUS and modified Menghini needle with adequate gauge were tools that enabled an effective, fast, safe, and low-cost front-line biopsy for patients with suspected LBCLs, avoiding in most cases psychological and physical pain of an unnecessary surgical intervention.

This study protocol was reviewed and approved by the Ethical Committee of AOU Federico II – Cardarelli: “Comitato Etico Campania 3,” via Pansini 5, 80131, Naples (Approval No. 66/2024).

The authors have no conflicts of interest to declare.

This study was not supported by any sponsor or funder.

M.P. and F.P. designed the research and performed the final revision of the manuscript. M.P., C.G., A.V., A.S., N.P., E.V., G.T., and M.M. performed the research, wrote the paper, and collected data. C.G., A.V., and N.P. analyzed data. C. Salvatore (Full Professor of Accounting and Business Administration, Department of Public Health, University of Naples Federico II) wrote the section devoted to cost analysis.

The data that support the findings of this study are not publicly available due to their information that could compromise the privacy of research participants but are available from the corresponding author (address email in the title page) upon reasonable request.