Utility of Bone Marrow Aspirate and Biopsy in Staging of Patients with T-Cell Lymphoma in the PET-Era – Tissue Remains the Issue

Abstract

The role of 18F-Fluoro-2-deoxy-D-glucose positron emission tomography combined with computerized tomography (PET-CT) in evaluation of bone marrow involvement (BMI) in patients with T-cell lymphoma (TCL) is poorly understood. We investigated whether PET-CT could replace bone marrow aspiration and biopsy (BMAB) in TCL. Sixty patients with newly diagnosed TCL who underwent both diagnostic PET-CT and BMAB were identified. BMI was tissue-confirmed in 15 (25%) cases, however only 8 of these 15 showed BMI on PET-CT (sensitivity of 53.3%, specificity of 100%). BMI by BMAB was associated with lower progression-free survival (PFS) (P=0.038) and overall survival (OS) (P=0.003) while PET-CT BMI was associated only with OS (P=0.02). BMI detected by BMAB in the setting of a negative PET-CT had similar inferior prognosis as BMI identified on PET-CT. Thus, PET-CT in TCL misses BMI in almost half of the cases detected by BMAB and hence cannot substitute BMAB in evaluation of TCL.

Introduction

Staging and treatment strategies in peripheral T cell lymphoma (PTCL) are guided by malignant involvement as detected by 18F-Fluoro-2-deoxy-D-glucose positron emission tomography combined with computerized tomography (PET-CT) imaging and bone marrow aspiration and biopsy (BMAB). Disease stage at diagnosis and international prognostic index (IPI) score are predictive of survival outcomes in PTCL. [1, 2] While around 10% of patients with diffuse large B cell lymphoma (DLBCL) will exhibit bone marrow involvement (BMI) at the time of diagnosis, [3] the incidence in TCL is higher at 25%; and these patients have an inferior OS. [4–6] Given this adverse prognosis, young and fit patients with advanced stage PTCL achieving first complete remission with a negative bone marrow at end of therapy are often offered consolidation with autologous stem cell transplantation. [7, 8]

In the modern era, PET-CT has evolved to be the preferred imaging modality for staging patients with newly diagnosed lymphoid malignancies. [9] Studies in DLBCL, other NHL, [10–14] and HL [15–17] have reported high diagnostic and prognostic performance, suggesting that PET-CT may replace BMAB for baseline bone marrow assessment. Based on these observations, the Lugano 2014 guidelines recommended that in HL and aggressive NHL, if PET-CT demonstrates focal uptake within the skeleton, then a BMAB is not required to prove involvement. [18, 19] However, these recommendations did not specify TCL patients. The majority of patients with TCL have FDG-avid/PET-positive disease, and compared to CT alone, PET appears particularly useful for detection of extra-nodal disease, suggesting that it should be routinely performed as part of initial evaluation of TCL. [20] However, evidence regarding the accuracy of PET-CT for detecting BMI in patients with TCL is lacking. In this study, we compared the performance of PET-CT and BMAB for detection of BMI in patients with newly diagnosed TCL.

Materials and Methods

Patients

We conducted a retrospective study of adult patients with newly diagnosed TCL treated at Roswell Park Comprehensive Cancer Centre (RPCCC) from 2004 to 2018. The study was approved by our institutional review board. Patients were identified through the cancer registry using the International Statistical Classification of Diseases and Related Health Problems (ICD-10) diagnosis codes; histologies included were peripheral T-cell lymphomas, not otherwise specified (PTCL-NOS), anaplastic large cell lymphoma (ALCL), angioimmunoblastic T-cell lymphoma (AITL) and natural killer T-cell lymphoma (NKTCL). Patients were excluded if either PET-CT and/or BMAB were missing at diagnosis or if BMAB specimens were diagnostically insufficient.

PET scan and Bone marrow biopsy

18F-Fluoro-2-deoxy-D-glucose PET/CT (Discovery LS, RX, 690 or10; General Electric, Waukesha, WI) was performed from skull vertex to the mid-thigh according to standard clinical practice on newly diagnosed TCL patients. PET was preceded by low dose helical CT for attenuation correction and anatomic orientation.

PET-CT data were independently reviewed by a board-certified nuclear medicine physician for confirmation of BMI. Prominent focal enhancement of radiopharmaceutical localization in bone marrow with an intensity exceeding that of normal liver and without an alternative explanation relating to non-oncologic disease, trauma, or procedural history was considered as baseline indication of BMI. Modest global elevation of deoxyglucose localization without focal lesions higher than normal liver background was not interpreted as BMI. [Figure 2] In all cases, BMAB was performed from the iliac crest at diagnosis and the site of biopsy was not guided by PET findings. Specimens were analysed by morphology, immunohistochemistry and flow cytometry for BMI, independent of the results of the PET-CT. Histopathological BMI was considered the gold standard in our study.

Figure 2. Bone marrow uptake on PET-CT.

2.A Focal metabolic lesions due to disease 2.B Global increase in marrow activity due to disease

Statistical Analysis

The diagnostic performance of PET-CT was summarized using the sensitivity (fraction of positive PET-CT in patients with confirmed BMI), specificity (fraction of negative PET-CT in patients with confirmed negative BMI), positive predictive value (PPV: fraction of confirmed BMI in patients with positive PET-CT), and negative predictive value (NPV: fraction with confirmed lack of BMI in patients with negative PET-CT). These measures were estimated with 95% confidence intervals obtained by Jeffrey’s prior method. Progression-free survival (PFS) was defined according to the 2007 International Working Group criteria [15] as time from diagnosis until disease progression, recurrence, or death from any cause. Overall survival (OS) was calculated from diagnosis until death or last follow-up. The survival outcomes (OS and PFS) were summarized using standard Kaplan-Meier methods, with comparisons made using the log-rank test. Univariate associations between survival outcomes and patient characteristics were evaluated using Cox regression models. Hazard ratios, with 95% confidence intervals, were obtained from model estimates. A covariate adjusted analysis was conducted using multivariable Cox regression models, where the associations between PET-CT or BMAB and survival outcomes were evaluated while adjusting for patient characteristics. Due to sample size constraints, the additional patient characteristics were only included one at a time. All analyses were conducted in SAS v9.4 (Cary, NC) at a significance level of 0.05.

Results

Patient characteristics

Eight seven TCL patients were initially identified, of which 27 patients were excluded due to either unavailability of a diagnostic PET-CT/BMAB or indeterminate results on the respective tests. A final cohort of 60 patients with TCL was analysed. Patient characteristics are summarized in Table 1. Median age was 58 years (range, 21–82) and 56% were male. The most common histological subtypes were PTCL-NOS (n = 21, 35%) followed by ALK- negative ALCL (n =15, 25%) and AITL (n = 11, 18.3%). Our cohort also included 7 patients (11.6%) with NKTCL. Chemotherapy regimens for PTCL subtypes were CHOP or CHOP-like in 39 patients (65%) and Hyper-CVAD in 11 patients (18.3%). Most NKTCL patients were treated with L-asparaginase or Hyper-CVAD based regimens. Median follow up was 92 months (range, 2 – 178). During this period, 33 patients (55%) developed progressive/relapsed disease and 34 patients (56.6%) died. The major causes of death were lymphoma progression (n=24), treatment related severe adverse events (n=3), secondary myeloid malignancy (n=1) and unknown in 6 patients.

Table 1.

Baseline Clinical Characteristics of Patients

Characteristic	Patients, n (%)
Age, median, years	58 (21–82)
Sex, male, n (%)	34 (56.6)
Histologic type
PTCL-NOS	21 (35)
AITL	11 (18.3)
ALCL – ALK+	6 (10)
ALCL – ALK -	15 (25)
NKTCL	7 (11.6)
ECOG performance status ≥ 2, n (%)	8 (13.3)
Ann Arbor disease stage ≥ 3, n (%)	41 (68.3)
Elevated serum LDH level	25 (41.6)
Extranodal lesions, ≥ 2, n (%)	11 (18.3)
IPI risk, n (%)
Low risk	24 (40)
Low-intermediate risk	17 (28.3)
High-intermediate risk	11 (18.3)
High risk	8 (13.3)
Bone marrow involvement on BMAB	15 (25)
Chemotherapy, n (%)
CHOP/ CHOEP based a	39 (65)
Hyper-CVAD	11 (18.3)
L-Asparginase based	4 (6.6)
Other regimens	6 (10)
Stem cell transplant recipients, n (%)
Autologous SCT	6 (10)
Allogeneic SCT	3 (5)

Abbreviations: PTCL-NOS, Peripheral T cell lymphoma- not otherwise specified; AITL, Angioimmunoblastic T cell lymphoma; ALCL, Anaplastic large cell lymphoma; ALK, Alkaline phosphatase; NKTCL, Natural Killer T cell Lymphoma; ECOG, Eastern Cooperative Oncology Group; LDH, Lactate Dehydrogenase; IPI, International Prognostic Index; BMAB, Bone marrow aspiration and biopsy; CHOP/CHOEP, cyclophoshamide, doxorubicin, vincristine, prednisone;, etoposide; Hyper-CVAD, Cyclophosphamide, vincristine sulfate, doxorubicin hydrochloride (adriamycin) and dexamethasone; SCT, stem cell transplant.

^aCHOP: n=34 (56.6%); CHOEP: n=5 (8.3%)

^bOther chemotherapy: Brentuximab Fludarabine, Gemcitabine, Cyclosporine, Romidepsin

Diagnostic performance of PET-CT

At diagnosis, 15/60 (25%) patients had lymphomatous BMI. While BMAB was positive in all 15 patients, PET-CT detected BMI only in 8 of these 15 patients (BM+/PET+; Sensitivity of 53.3%) [Table 2]. In the remaining 7 patients, PET-CT was classified negative despite a positive BMAB (BM+/PET-). The clinical characteristics, histopathological and imaging features of these patients are summarized in Table 3. In our cohort, 13/60 (21.6%) patients had diffusely increased BM FDG uptake on PET-CT without focal lesions. Per the aforementioned definition, these were documented but not interpreted as BMI by PET-CT. None of the patients in our cohort had a positive PET-CT for BMI with a negative BMAB (BM-/PET+). Concordance between PET-CT and BMAB was noted in 53 out of the 60 patients, 45 with BM-/PET- and 8 with BM+/PET+ [Table 2]. Accordingly, the specificity, PPV and NPV of PET-CT in identifying BMI was 100%, 100% and 86.5% respectively.

Table 2.

Comparison of ability of BMAB and PET-CT to detect Bone Marrow Involvement

	Bone Marrow Biopsy
	Positive	Negative
PET-CT positive	8	0	PPV 100% (95% CI 73.8, 100)
PET-CT negative	7	45	NPV 86.5% (95% CI 75.4, 93.8)
	Sensitivity 53.3% (95% CI 29.4, 76.1)	Specificity 100% (95% CI 94.6, 100)

Abbreviations: PPV, positive predictive value; NPV, negative predictive value, CI, Confidence interval

Table 3:

Clinical characteristics, histopathological and imaging features of patients with biopsy proven bone marrow involvement in the setting of a negative PET-CT

Clinical Features	Diagnosis	Histopathological features on bone marrow biopsy	Bone marrow findings on initial PET-CT
72y female with a mediastinal mass lymphocytosis, pruritus fatigue	PTCL-NOS	Extensive infiltrate of CD3+, CD5+clonal T cell lymphocytes involving 25 % of the marrow with advanced trilineage hematopoiesis.	Uniform marrow uptake equivalent to liver & spleen without focal variations. Soft tissues lesions are twice more intense than global bone activity
66y female with B symptoms and generalized lymphadenopathy	AITL	Extensive CD3+, CD4+ clonal T lymphoid infiltrate involving 25 % of cellularity with myelofibrosis and polyclonal plasma cell proliferation.	Diffuse uniform bone marrow uptake equivalent to liver & spleen without focal variations. Marrow activity is significantly lower than soft tissue lesions
70y female with B symptoms and extensive lymphadenopathy	AITL	Hypercellular BM with myelofibrosis, granulocytic hyperplasia, eosinophilia. CD3+ clonal T lymphoid aggregates consistent with minimal involvement by T cell lymphoma	Diffuse heterogenous bone marrow activity slightly greater than liver but without focal lesions. Soft tissue lesions are four times more intense than global bone marrow activity
58y female with B symptoms and pancytopenia	PTCL-NOS	Markedly hypercellular marrow. Flow cytometric and immunohistochemical findings consistent with mature T cell leukemia/lymphoma comprising 30–40% of the cellularity	Mild generalized metabolic uptake in the skeletal system which does not exceed that of the liver metabolic uptake. No focal lesions identified
44y male with B symptoms, pancytopenia and rash	ALCL	Multifocal nodular CD2+, CD3+, CD4+, CD5+ clonal T lymphoid proliferation occupying 15% of cellularity with granulocytic hyperplasia, eosinophilia with grade 2 reticulin myelofibrosis.	Generalized metabolic uptake in the marrow not exceeding liver metabolic uptake. No focal lesions noted.
47y male with B symptoms, mediastinal mass and extensive adenopathy	PTCL-NOS	Normocellular marrow; Few CD3+ T-cell rich lymphoid aggregates with focal PD1 positivity and clonal T cell receptor gene rearrangement, consistent with minimal involvement by T-cell lymphoma	Uniform marrow uptake equivalent to liver & spleen without focal variations. Soft tissues lesions are four times more intense than global bone activity.
57y male with severe anemia and hepato-splenomegaly	AITL	Hypercellular marrow, erythroid dysplasia; 5 % involvement with CD3+, CD4+, CD52+ angioimmunoblastic T cell lymphoma	Uniform marrow uptake slightly greater than liver and equivalent to spleen but without focal lesions. Soft tissue lesions are twice as intense as global bone marrow activity

Abbreviations: PTCL-NOS, Peripheral T cell Lymphoma-Not Other Specified; AITL, Angioimmunoblastic T cell lymphoma; ALCL, Anaplastic large cell lymphoma, BM, bone marrow, y, year

Effect of either modality on staging was evaluated by computing the Ann Arbor stage using PET-CT with and without BMAB results. BMAB upstaged 2/32 (5.7%) patients with Stage I/II TCL to stage IV due to identification of BMI otherwise not detected by PET-CT. [Supplementary Table 1]

Survival

Median PFS and OS for the entire cohort was 28 (range, 11–51) months and 76 (37 −139) months respectively. Patients with BMI detected by PET-CT had a significantly shorter OS (3yr OS 25% vs 70%; p=0.02), however, not PFS as compared to those without BMI on PET-CT (3yr PFS 25% vs 45%; p=0.26) [Figure 1.A]. In comparison, both PFS and OS were significantly lower in patients with BMI by BMAB compared to those without BMI by BMAB (3yr PFS 27% vs 48%; p=0.038) and (3yr OS 33% vs 74%; P=0.003) respectively [Figure 1.B]. Patients with BM+/PET- BMI had a significantly lower OS when compared to patients with BM-/PET- BMI (3yr PFS 33% vs 51%; P=0.065) and (3yr OS 50 % vs 72%; P=0.041) respectively. Furthermore, outcomes in patients with BM+/PET- BMI were similar to patients with BM+/PET+ BMI [3yr PFS 29% vs 25%; p=0.64) and (3yr OS 43 % vs 25%; p=0.85) respectively [Figure 1.C].

Figure 1. Kaplan Meier survival curves of bone marrow involvement detected by PET-CT and bone marrow biopsy in TCL patients.

A Bone involvement detected on PET-CT was associated with a worse OS but not PFS. B Bone involvement detected on BMAB was associated with a worse PFS and OS. C BM+/PET- had a significantly lower OS when compared to patients with BM-/PET-. Outcomes in patients with BM+/PET- were similar to patients with BM+/PET+

Cox regression models were used to evaluate the association of PFS and OS with demographic or clinical characteristics such as age ( >60 vs less than 60 years), elevated LDH, stage (I-II vs III-IV) , ECOG performance status (0–1 vs 2–4), extranodal sites and high IPI score (0–2 vs 3–5). High IPI score was associated with lower PFS and OS and presence of extra-nodal disease was associated with inferior OS [Supplementary Table 2]. The covariate adjusted analyses indicate that BMI by PET-CT and BMAB maintained their prognostic impact on survival outcomes when accounting for each of the above-mentioned clinical factors [Supplementary Table 3-4].

Discussion:

In the present study, we sought to determine the role of PET-CT in the initial evaluation of lymphomatous bone marrow involvement in patients with TCL and whether it is sufficiently accurate to render a BMAB unnecessary in the staging of TCL. We used positive BMAB as the standard criterion for estimation of BMI. In our cohort, while the specificity was 100%, the sensitivity of PET-CT to detect BMI was only 53.3%. The prognosis of patients with BMAB+/PET- was as poor as those with a positive PET-CT suggesting that this cohort of patients need to be accurately identified and treated aggressively. We did not have a single patient with BMAB negative but PET-CT positive for BMI. It is conceivable that this variation from other reported studies may be a result of varying standards of reference. In prior studies ²², diffuse FDG uptake in the BM was considered to represent BMI. Moreover, PET-CT findings were often not confirmed by biopsy. In our study, CT images of the combined PET/CT scan were carefully reviewed and those with diffuse rise in uptake without focal lesions significantly higher than normal liver background were designated as modest increase or expansion of marrow activity that was possibly reactive in nature, and not BMI. PET-CT was rated positive for BMI only if FDG uptake was focally variant displaying an intensity exceeding that of the normal liver not attributable to benign cause, conceivably imparting greater specificity for image-based interpretation of BMI. PET-CT in our study had a high false negative rate of 46% for BMI. Although the negative predictive value of PET-CT in ruling out BMI seems high at 88.5%, this may be due to the small number of patients with a positive BMI (16/60 patients). On the other hand, using BMAB as a modality for detection of BMI, at least 7 patients with stage I – III disease on initial PET-CT were upstaged to Stage IV disease and this was associated with a worse PFS and OS.

Previous retrospective studies have reported similar results with high specificity but relatively low sensitivity of PET-CT to detect BMI in TCL. El-Galaly et al assessed the utility of interim and end-of-treatment PET-CT in 114 cases of PTCL. [21] In their study, staging PET-CT at diagnosis had a sensitivity of only 18% with a specificity of 90% to detect biopsy-proven BMI. The presence of focal skeletal PET-CT avid lesions did not predict for a worse PFS or OS, however, a positive BMAB was associated with an inferior PFS (HR 2.03, 95% CI 1.08–3.84) and OS (HR 2.42, 95% CI 1.19–4.93). Koh et al [22] evaluated 109 patients with PTCL (n=63) and extra nodal NK/T-cell lymphoma (n=46). The sensitivity of PET-CT in detecting BMI was 64%. Unlike our study, this study included a significant proportion of patients with NK-TCL, which has a peculiar disease distribution and is usually associated with a low frequency of BMI.

In contrast, Abe and colleagues [23] studied a large population of 83 patients with PTCL and reported the diagnostic performance of PET-CT for BMI to be better than BMAB (sensitivity, 89.3% vs 60.7%) with a specificity of 100%. In the same study, PET-CT missed biopsy proven BMI in 3 of 28 patients (10.7%) suggesting that PET-CT cannot completely replace BMAB. The reference standard for true BMI in their study was not only a positive BMAB but also included focal skeletal lesions that subsequently disappeared with therapy. It is important to note that TCL can frequently induce secondary changes in the marrow microenvironment independent of lymphomatous BMI, such as granulocytic hyperplasia, eosinophilia, granulomatous inflammation, plasmacytosis, vascular proliferation [24] and rarely hemophagocytosis. [25] Histologies such as AITL are particularly associated with such marked polymorphous cell population that may produce different metabolic phenotypes on PET-CT. [24] A notable point is that 33% of the cohort this study [23] comprised of AITL patients and this may have potentially led to overestimation of the diagnostic performance of PET-CT.

Our study is limited by its retrospective nature and small number of patients. Frontline treatments were heterogenous with some patients receiving more aggressive regimens such as Hyper-CVAD that may have caused an unexpected bias in survival outcomes. However, the percentage of patients with BMI was fairly balanced across treatment groups. PET-CT equipment, image acquisition, processing and interpretation criteria have evolved over the years and hence, interobserver variation may have altered the reliability of PET-CT interpretation. To obviate that shortcoming, an independent review all PET images was conducted by a nuclear medicine physician with extensive experience spanning decades in the evaluation of lymphoreticular neoplasms using FDG-PET; this led to revision of PET-CT interpretation of BMI in 7/60 patients (11.6%). Characteristic imaging findings that led to revision of the PET-CT on re-interpretation in these patients are tabulated in Supplementary Table 5. It is important to note that many of the published series of PET for assessment of BM status are reported from academic tertiary referral centres with high scan volumes and considerable reporting expertise; hence the generalizability of PET-CT findings remains uncertain. NKTCL comprised of 11.6 % (n=7) of our total cohort. Given the low incidence of BMI in this population (only 1 out of the 7 patients) and possibility of underestimation of diagnostic performance of PET-CT, we performed a subgroup analysis in PTCL patients, excluding NKTCL histology. We found the performance of PET-CT and BMAB in detection of BMI and their respective impact on prognosis to be similar in this cohort as compared to that seen above in the whole population of TCL. [Supplementary Figures]

Based on these findings, we advocate for the following approach to evaluate BMI in TCL. If PET-CT shows features of focal abnormal bone marrow uptake, a BMAB may be avoided as there is high probability of lymphomatous BMI based on the high specificity and positive predictive value for PET-CT noted in our study and others. This approach will be especially useful in patients with conditions like obesity or osteoporosis that make it technically challenging to perform a BMAB. We recommend BMAB evaluation in situations of diffuse BM FDG uptake higher than liver as these may be related to secondary hematological changes described above. When PET-CT shows no evidence of BMI, a BMAB should be considered based on clinical data and /or when presence or absence of BMI may significantly affect the IPI risk categorization [Figure 3].

Figure 3.

Algorithmic approach to bone marrow evaluation in the initial staging of T cell lymphoma

In conclusion, the performance of PET-CT in identification of BMI in TCL is fraught with a high rate of false negative results and patients should not routinely be assumed to have negative BMI based solely on PET-CT. While upstaging patients with lymphomatous BMI on BMAB did not impact treatment decisions in our study, it did provide robust prognostic information. Therefore, BMAB should be considered in all patients with newly diagnosed TCL and an algorithmic approach adopted to determine when it might be safe to omit a BMAB in an individual patient.