Can clinical data help to screen patients with lymphoma for MR imaging of bone marrow?

Annals of Oncology 6: 795-800, 1995. © 1995 Klicwer Academic Publishers. Printed in the Netherlands.

Original article Can clinical data help to screen patients with lymphoma for MR imaging of bone marrow? A. A. Tardivon,1 J.-N. Munck,2 L. G. Shapeero,1'6 S. Koscielny,3 J. Bosq,4 F. Dhermain,5 R. Gilles,1 M. Hayat2 & D. Vanel1 Departments of l Radiology, 2 Hematology, ^Statistics, * Pathology, !Radiotherapy, Institut Gustave Roussy, Villejuif, France; 6 Department of Radiology University of California, San Francisco, CA, U.S.A.

Background: Previous studies have suggested combining magnetic resonance (MR) imaging and biopsy in patients with lymphoma but associations between MR results and clinical symptoms have never been investigated. The purpose of this retrospective study was to better delineate patients profiles requiring bone marrow (BM) imaging in lymphoma. Material and methods: 50 MR studies and blind biopsies (BB) were reviewed in 40 patients with lymphoma. MR results were compared to clinical, laboratory-based and BM follow-up data to determine potential associations between MR results and these parameters. Results: 46% of MR studies were abnormal with a normal

Introduction

Knowledge of the bone marrow status is essential for the staging and treatment of patients with malignant lymphoma [1]. In Hodgkin's disease (HD), it contributes to the Ann Arbor classification which is critical for the therapeutic strategy: localized disease without bone marrow involvement is mainly treated with radiotherapy [2]. For aggressive non-Hodgkin's lymphoma (NHL), the stage will indicate to which risk group patients should be allocated for treatment with chemotherapy of varying intensity [3-4]. The bone marrow status is of further import in patients who relapse since most of them will be treated with therapy including autologous haematopoeitic cell transplantation [5-6]. Unilateral bone marrow aspiration and/or biopsy, usually of the posterior iliac crest, is the primary means of assessing the bone marrow status [7-9]. Bone marrow infiltration is found in 5%-15% of patients with HD and 25%-40% of patients with NHL [1]. Unilateral blind biopsy (BB) is subject to frequent sampling errors compared to bilateral biopsies [10,11]. Moreover, because the infiltrated bone marrow frequently exhibits a focal pattern of distribution in most cases of HD and high grade NHL, bone marrow involvement is also likely to be missed at bilateral BB. Magnetic resonance (MR) imaging is able to assess

BB; 2% were normal with an abnormal BB. Abnormal MR results were significantly associated with subsequent bone marrow involvement (p<0.01). Abnormal MR studies were significantly associated with constitutional symptoms, bone pains (p<0.05) and an elevated alkaline phosphatase level (p < 0.01). MR imaging excluded malignancy in three patients and caused therapy to be modified in three. Conclusion: Abnormal clinical and laboratory-based data should be used to screen patients with normal BB for MR imaging, especially in patients with Hodgkin's disease and high grade non-Hodgkin's lymphoma. Key words: biopsy, bone marrow, comparative studies, lymphoma, magnetic resonance studies

the entire bone marrow compartment and has been shown to be the most useful technique for the imaging of bone marrow disorders [12-19]. Previous studies have suggested combining MR imaging and biopsy in patients with lymphoma but associations between MR results and clinical symptoms have never been investigated [17-19]. In 40 patients with lymphoma, 50 MR studies and BB were first compared to highlight discrepancies between their respective findings. MR imaging was then compared to clinical and laboratorybased data to search for potential associations between MR imaging and these parameters which would better define patient profiles requiring bone marrow imaging. Material and methods Population Bone marrow MR imaging (50 studies), BB results and pertinent clinical data were obtained in 40 consecutive patients with malignant lymphoma. Patient characteristics are presented in Table 1. Thirty-three patients had a single MR study, five had 2 MR studies, one patient had 3 MR studies and another one had 4 MR studies. A unilateral blind bone marrow biopsy of the posterior iliac crest was performed within a mean interval of 11 days from the MR study. All BB werereviewedby the same pathologist. The following clinical and laboratory-based parameters were evaluated at the time of each MR study: the clinical status of the

Downloaded from https://academic.oup.com/annonc/article-abstract/6/8/795/204333 by guest on 24 October 2019

Summary

796 Table 1. Repartition of the patients according to their characteristics and disease. 40

Hodgkin's disease Initial presentation Relapse Histology Nodular sclerosis Mixed cellularity Lymphocyte depletion

26 12 14 17 7 2

Non-Hodgkin's lymphoma Initial presentation Relapse Histology Low grade Intermediate grade High grade

14 8 6

Statistical analysis

Stage I—II

m-rv No evidence of disease with clinical or lab abnormalities Constitutional symptoms A B Follow-up — bone marrow evolution pattern Normal Abnormal

24 16 8 32

2 6 6 5

36

Fisher's exact test was used to test the associations between MR results and clinical and laboratory-based data [20], and between MR and bone marrow follow-up results; the sensitivity and specificity of MR imaging were calculated taking into account follow-up data on bone marrow status. A p value of less than 0.05 was considered significant.

9 23 17

Results MR imaging

21 19

Thirty-one MR studies with T, -weighted images were abnormal. The site of MR bone marrow abnormalities Follow-up data was: the spine alone - 22%, the pelvis alone - 9%, the Alive 35 Complete remission 19 femurs alone = 3%, spine and pelvis together — 25% Partial remission 16 and all the three regions - 41%. Bone marrow involveDeath (specific) 5 ment was nodular in 65% of patients with HD and in 85% of high grade NHL. Nineteen T,-weighted MR studies were normal. lymphoma, and in patients with disease, stage according to the Ann Of the T2 or T2*-weighted images (37 MR studies), Arbor classification (without taking into account MR data), the 13 showed bone marrow abnormalities. Nine of these presence of bone pain, the erythrocyte sedimentation rate (ESR), alkaline phosphatase and lactate deshydrogenase levels, haemo- 13 studies (69%) exhibited the same number of lesions globin concentration and leukocyte, granulocyte and platelet counts. as Tl -weighted images (eight cases of nodular involveBone marrow scintigraphy and data from plain radiographs were not ment and one case of diffuse involvement) and 4/13 included because the number of studies were limited. (31%) showed fewer lesions than T[ -weighted images The mean follow-up from the time of the MR study was 23 months. When the follow-up of the patients was taken into account, (small-sized nodular involvement with focal or diffuse evidence of subsequent bone marrow involvement was considered distribution). On the other hand, 24 T 2 or T*-weighted to indicate that previous infiltration, undetected at histology, was in studies were normal: 14 were concordant with T,fact present During the follow-up, patients with one of the following weighted studies whereas 10 were abnormal on T, parameters were considered to have a positive bone marrow status: weighted images (seven cases of diffuse involvement, positive BB (n —6), positive surgical biopsy (n-2), positive bone scintigraphy (n-2), positive computed tomography study (n-2), one of nodular involvement with diffuse distribution changes on MR imaging (an increase in size of lesions — 6 and a de- and two of nodular involvement with focal distribucrease in size of lesions — 3). tion). Because Tj -weighted MR imaging was performed in all the patients and provided more bone marrow abnorMR studies malities then T2 or T*-weighted studies, only the TlMR imaging of bone marrow was performed on a 1.5 Tesla system weighted results were used for the analysis with BB, (General Electric Medical Systems, Milwaukee). Routinely, the clinical and laboratory-based data. dorsolumbar spine, the pelvis and both proximal femurs were investigated with contiguous coronal and sagittal images with 5 mm section thickness. In two patients, the proximal extremity of tibias

Downloaded from https://academic.oup.com/annonc/article-abstract/6/8/795/204333 by guest on 24 October 2019

Number of patients Age: range (15-58), mean (32) <30 >30 Sex Female Male

was also imaged. T, -weighted spin-echo sequences were acquired in all the patients (450-500 msec/12-20 msec, matrix of 256 x 192 or 256 x 256, two excitations). T2-weighted spin-echo sequences (1800-2000 msec/25-100 msec, matrix 256 x 128, one excitation) were obtained for 19 MR studies and gradient-echo sequences (500 msec/12 msec/110') for 18 MR studies. Fifty MR studies were examined by a senior radiologist for the presence of bone marrow involvement without knowledge of clinical data nor of the histopathologic diagnosis; T, and T2 or TJ-weighted images were analyzed together. The bone marrow was considered positive on the basis of the following previously reported criteria [16-19]: on T,-weighted MR images: 1) a diffuse homogeneous decrease in high signal intensity of fatty marrow, 2) multiple nodular areas of focal low signal intensity. Diffuse or focal distribution of MR abnormalities was also noted so that discordant results between MR imaging and BB findings could be analyzed. On T2 or TJweighted images: 3) an increase in bone marrow signal intensity at the same site of T, -weighted abnormalities.

797 Comparison MR imaging - blind biopsy (Table 2)

Table 2. T,-weighted MR imaging versus BB and bone marrow evolution pattern. n - 50 studies Abnormal MR study Normal MR study

Abnormal MR study Normal MR study

Abnormal BB

Normal BB

8 (16%) 1 (2%)

23 (46%) 18 (36%)

Positive follow-up

Negative follow-up

28 (56%) 1 (2%)

3 (6%) 18 (36%)

Fig. 2. Pelvic bones on coronal T,-weighted image. Nodular involvement of iliac bone marrow with diffuse distribution. The blind biopsy was negative (Hodgkin's disease).

a normal MR study and normal BB (18 studies) did not develop bone marrow infiltration subsequently, but seven of them (39%) were treated after the MR study. The sensitivity and specificity of MR imaging were 96.5% and 86%, respectively, with an overall accuracy of 82%. Comparison MR imaging - clinical data (Table 3) An abnormal MR study was significantly associated with the presence of constitutional symptoms or bone pain or an elevated alkaline phosphatase level. No significant association was found between MR imaging and the other parameters, in particular, with the clinical stage of lymphoma, the lactate deshydrogenase level, the ESR or blood counts. Only one patient with constitutional symptoms and a normal MR study had an abnormal biopsy (Fig. 1). Two of the three patients with bone pain had benign abnormalities on MR studies: one patient had benign vertebral fractures (Fig. 3) and the other avascular necrosis of the femoral head. The third case remains in clinical remission with isolated bone pain. Constitutional symptoms, bone pains and an elevated alkaline phosphatase level were not found together in any of these cases. MR imaging and the therapeutic decision

Due to abnormal MR results, the therapeutic strategy was modified in three patients with HD. In two of them, the radiation fields were modified and included the femurs in the first one (Fig. 4) and the 8th thoracic vertebra in the second one. The third patient had previously been treated with chemotherapy and subtotal irradiation (initial stage HI). After a clinical remission of nine months, he developed bone pains, fever and an Fig. 1. False-negative MR study (posterior iliac crests on coronal T, -weighted image). Blind biopsy was abnormal in this patient with elevated ESR with a normal BB. The isolated bone intermediate NHL. No bone marrow involvement was found in iliac marrow relapse was detected by MR imaging and the bones on MR study. patient received chemotherapy. After the first cycle of

Downloaded from https://academic.oup.com/annonc/article-abstract/6/8/795/204333 by guest on 24 October 2019

Concordant normal results were obtained in 36%, and concordant abnormal results in 16%. Discordant results were found in 48% of cases. The BB result was abnormal and MR study normal in only one patient (diffuse, small cleaved cell malignant lymphoma) with general symptoms and bone pain (Fig. 1). On the other hand, BB was normal in 23 abnormal T, -weighted MR studies in 18 patients (8 initial disease and 10 relapse). Pelvic bones were examined in these 23 abnormal MR studies because all the normal biopsies had been performed in this area. No infiltration was found in seven MR studies (30%); bone marrow infiltration was nodular with focal distribution in five (22%), nodular with diffuse distribution in five cases (22%) (Fig. 2) and diffuse alone in six MR studies (26%). When the follow-up of the patients was taken into account (Table 2), the positivity of MR imaging was significantly associated with subsequent bone marrow infiltration during the follow-up (p < 0.01). Among the 18 patients with an abnormal MR and a normal BB (23 studies), only three remained free of bone marrow involvement during their follow-up. The 18 patients with

798 Table 3. Associations between MR results, clinical and laboratorybased data (ns - non significant). Variables

Stage Stage I-n

stage m-rv

% of abnor- Fischer's mal MR exact test studies p value

5

80% 69%

ns ns

38

12

53% 92%

p < 0.05

21 23

44% 83%

p < 0.01

li 31

50% 68%

ns

32 13

56% 69%

ns

22 25

41% 76%

p < 0.05

34 14

62% 57%

ns

6 42

50% 62%

ns

9 39

67% 59%

ns

43 6

60% 67%

ns

36

Fig. 3. Lumbar spine on T, -weighted image (treated intermediate NHL in clinical remission with dorsolumbar pain). MR imaging revealed benign vertebral fractures (white arrows). Note the posttherapeutic band pattern (black arrow) of peripheral intermediate signal intensity.

treatment, bone pain and constitutional symptoms disappeared and the ESR normalized. At the end of chemotherapy, MR imaging was normal. Discussion The major findings in our study are: 1) more abnormalities suggestive of malignant infiltration were found with T, -weighted bone marrow imaging than with T2 or T*-weighted imaging; 2) discordant results were found in 48% of cases between MR results and BB findings and follow-up data reinforced the value of MR abnormalities as predictive of bone marrow infiltration; 3) the presence of constitutional symptoms or bone pain or an elevated alkaline phosphatase level was significantly associated with an abnormal MR study. T,-weighted images provided more abnormalities suggestive of bone marrow infiltration than T2 or T*weighted images, especially when infiltration was diffuse or nodular with small-sized nodules. Previous studies have shown a wide variability in T2 relaxation times due to the nature of the different lymphomatous lesions [12,17]. The low signal/noise ratio of T 2 -

Fig. 4. Therapeutic impact of bone marrow imaging (coronal T, weighted image). There was obvious nodular bone marrow infiltration of the two proximal femurs. Radiation fields were modified to include these segments (patient with Hodgkin's disease).

weighted scans, the association of myelofibrosis and the magnetic suceptibility artifacts inherent to T 2 weighted images could account for these discordant findings. One of the limitations of our study was the absence of fat suppression sequences known to be more sensitive for the detection of bone marrow infiltration, especially when T,-weighted images were abnormal and their T2-weighted counterparts normal. However,

Downloaded from https://academic.oup.com/annonc/article-abstract/6/8/795/204333 by guest on 24 October 2019

Constitutional symptoms Without - A With-B Bone pain Without With Erythrocyte sedimentation rate (ESR) <30 >30 Lactate deshydrogenase level Normal Elevated Alkaline phosphatase level Normal Elevated Hemoglobin concentration >10g/100 mL <10g/100 mL Leukocyte count <2000/mm 3 >2000/mm 3 Granulocyte count <2000/mm 3 >2000/mm 3 Platelet count <400 000/mm 3 >400 000/mm 3

No. of MR studies

799 with bone marrow infiltration assessed by MR as well as by bone marrow follow-up in this small population. Other values such as the ESR, the lactate deshydrogenase level and blood counts had no predictive value, and this could be due to interferences caused by treatment or to the small sample studied. Conclusion Our results suggest a critical role for MR in bone marrow assessment of patients with Hodgkin's disease and aggressive lymphoma with normal BB and clinical parameters associated with bone marrow involvement. Such parameters have been determined in our retrospective study but prospective evaluation, with multivariate analysis of a larger population, is warranted. Acknowledgment We thank Lorna Saint Ange for editing this manuscript. References 1. McKenna RW, Hernandez JA. Bone marrow in malignant lymphoma. Hematol/Oncol Clin North Am 1988; 2(4): 617-35. 2. HartseU WF, Sarin P, Recine DC et al. Long-term results of curative irradiation in pathologically staged IA and IIA Hodgkin disease. Radiology 1993; 186: 565-8. 3. Coiffier B, Gisselbrecht C, Vose JM et al. Prognostic factors in aggressive malignant lymphomas: Description and validation of a prognostic index that could identify patients requiring a more intensive therapy. J Clin Oncol 1991; 9: 211 -9. 4. The International Non-Hodgkin's Lymphoma Prognostic Factors Project A predictive model for aggressive non Hodgkin's lymphoma. N Engl J Med 1993; 329: 987-94. 5. Me Millan AK, Goldstone AH. Autologous bone marrow transplantation for non-Hodgkin's lymphoma. Eur J Haematol 1991; 46:129-35. 6. Majolino I, Quaglietta AM, Iacone A et al. Autologous blood stem cell transplantation in malignant lymphomas: An Italian cooperative study. Leukemia and Lymphoma 1992; 7(S): 11-6. 7. Bartl R, Frisch B, Burkhardt R et al. Lymphoproliferations in the bone marrow: Identification and evolution, classification and staging. J Clin Pathol 1984; 37: 233-54. 8. Bartl R, Frisch B, Burkhardt R et al. Assessment of marrow trephine biopsy in relation to staging in chronic lymphocytic leukaemia. Br J Haematol 1982; 51:1-15. 9. Westerman MP. Bone marrow needle biopsy: An evaluation and critique. Semin Hematol 1981; 18: 293-300. 10. Brunning RD, Bloomfield DC, McKenna RW et al. Bilateral trephine bone marrow biopsies in lymphoma and other neoplastic disorders. Ann Intern Med 1975; 82: 365-6. 11. Coller BS, Chabner BA, Gralnick HR. Frequencies and patterns of bone marrow involvement in non-Hodgkin's lymphomas: Observations on the value of bilateral biopsies. Am J Hematol 1977; 3: 105-19. 12. Daffner RH, Lupetin AR, Dash N et al. MRI in the detection of malignant infiltration of bone marrow. AJR 1986; 146: 353-8. 13. Nyman R, Rehn S, Glimelius B et al. Magnetic resonance imaging in diffuse malignant bone marrow diseases. Acta Radiol 1987; 28(2): 199-205. 14. Vogler JB HI, Murphy WA. Bone marrow imaging. Radiology 1988; 168: 679-93.

Downloaded from https://academic.oup.com/annonc/article-abstract/6/8/795/204333 by guest on 24 October 2019

the follow-up data confirmed bone marrow infiltration in these cases. MR imaging was more sensitive than blind biopsy which is consistent with previous studies [17-19]. In the literature, the percentage of abnormal MR studies, using standard spin echo sequences ranged from 10% to 20%, when BB was normal [17-19]. In our study, various factors account for the 46% of abnormal MR studies with normal BB. A significant proportion of our patients had MR studies because their BB was normal. In 74% of patients, pelvic bone marrow was normal or had undergone malignant nodular infiltration which explains the large proportion of normal biopsies. In addition, 56% of the studies were obtained in patients who relapsed and BB was often hypocellular because of previous chemotherapy and/or radiation. One of the major problems in our study was to accurately establish the bone marrow status. The best way to correlate MR abnormalities with bone marrow involvement would have been to perform guided biopsies, but this was not attempted in this study. Previous reports, in which MR imaging was used to assess the bone marrow status, have reported only a few cases confirmed by guided biopsies [18,19]. Here, with an appropriate follow-up for all the patients, detection of subsequent bone marrow infiltration failed in only three patients. In the 18 patients with a normal MR study, a normal BB and negative bone marrow status, seven of them were treated after the MR study. It cannot be excluded that false negative MR studies were masked as a result of effective treatment. MR imaging failed to detect malignant infiltration in one patient with a diffuse, small cleaved cell malignant lymphoma but the biopsy was positive. This limitation of MR imaging has been described in patients with low grade lymphomas [13,16]. In such cases, the quantitative assessment of bone marrow signal intensity could be useful in establishing the diagnosis of bone marrow infiltration in the few cases in which BB are negative [17]. Overall, our study confirms the value of MR as a useful tool to predict bone marrow involvement in lymphoma. However, it is a costly, cumbersome investigation and its systematic use must be excluded. Furthermore, its role in the assessment of the bone marrow during the work-up in lymphoma has to be redefined. As already mentioned, it probably is of limited interest in lymphocytic lymphomas and possibly other low grade lymphomas, where BB are abnormal most of the time. In high and intermediate grade lymphomas as well as in Hodgkin's disease, MR imaging will have a role to play when BB is normal. Its indication will probably have to be limited to situations in which there is a high probability of bone marrow involvement since many of these patients have no marrow sites of disease, and indicators such as those described in this study may help to refine patient selection. The presence of bone pain, constitutional symptoms and an elevated alkaline phosphatase level were strongly associated

800 15. Kattapuram SV, Khurana, Scott JA, El-Khoury GY. Negative scintigraphy with positive magnetic resonance in bone metastases Skeletal Radiol 1990; 19: 113-6. 16. Algra PR, Bloem JL, Tissing H et al. Detection of vertebral metastases: Comparison between MR imaging and bone scintigraphy. Radiographics 1991; 11: 219-32. 17. Dohner H, Giickel F, Knauf W et al. Magnetic resonance imaging of bone marrow in lymphoproliferative disorders: Correlation with bone marrow biopsy. Br J Haematol 1989; 73: 12-7. 18. Linden A, Zankovich R, Theissen P et al. Malignant lymphoma: Bone marrow imaging versus biopsy. Radiology 1989; 173: 335-9. 19. Shields AF, Porter BA, Churchley S et al. The detection of

bone marrow involvement by lymphoma using magnetic resonance imaging. J Clin Oncol 1987; 5(2): 225-30. 20. Kendall M, Stuart A. The Advanced Theory of Statistics, Vol 2. New York: MacMillan Publishing Company, Drc. 1979. Received 22 February 1995; accepted 5 July 1995. Correspondence to: Anne A. Tardivon, M.D. Department of Radiology Institut Gustave Roussy 94805 VillejuifCedex France

Downloaded from https://academic.oup.com/annonc/article-abstract/6/8/795/204333 by guest on 24 October 2019