Fluorodeoxyglucose positron emission tomography/computed tomography for evaluation of bone marrow involvement in lymphoma: when is it superior to biopsy?

Bone marrow biopsy (BMB) is widely performed in patients with either Hodgkin lymphoma (HL) or non-Hodgkin lymphomas (NHLs) to provide diagnostic and prognostic information. Bone marrow involvement (BMI) upstages patients to stage IV disease and may direct more intensive therapeutic strategies. It is, however, an invasive procedure frequently associated with anxiety and pain, and has a small risk of hemorrhage or sedation-related complications [1,2]. It is also resource-intensive with the potential to delay commencement of therapy. It is performed blindly, usually in the posterior iliac crest, and assesses only a small portion of the entire marrow based on the assumption that tumor cells spread diffusely through marrow. Accordingly, it has high specificity but potential for low sensitivity due to sampling error if BMI is focal and not diffuse. This is highlighted by studies demonstrating significant discordance of results in patients undergoing bilateral BMB, with discrepancy rates as high as 60%, or higher detection rates with longer trephine length [3–5].

Imaging is increasingly playing a central role in the management of patients with lymphoma. Its use has led to improved outcomes through more accurate staging and also reduced morbidity through minimization of invasive procedures. Use of computed tomography (CT) in the 1970s replaced the need for staging laparotomy and invasive lymphangiography for nodal staging [6]. Nuclear medicine with gallium-67 provided the template for functional imaging, particularly for restaging, as it allows differentiation of active lymphoma from residual fibrotic tissue, which is difficult with anatomical imaging [7]. Positron emission tomography (PET) with fluorodeoxyglucose (FDG) emerged in the 1980s as a sensitive tumor-detecting radiopharmaceutical demonstrating high management impact in patients with lymphoma [8,9]. FDG PET has provided new insights into the extent and distribution of BMI by demonstrating focal marrow involvement in some patients, thereby highlighting the weakness of blind biopsy.

Early studies suggested that FDG PET had inferior sensitivity for the detection of marrow involvement compared to BMB and, whilst complementary in some patients, could not replace BMB. These studies were limited by the use of stand-alone PET technology and analysis of heterogeneous cohorts, including patients with low and high grade disease, some of whom had already undergone therapy [10]. Some early studies using blinded BMB as the gold-standard spuriously suggested false-positive PET results, but when directed biopsies of FDG-avid bone abnormalities were performed, all revealed lymphomatous infiltration [11]. Hybrid PET/CT improved accuracy by enabling precise anatomical characterization of any functional abnormality [12,13] and also enabled rapid attenuation correction, which is particularly valuable in the assessment of marrow. Even in the PET/CT era, there have been numerous advances, including use of newer-generation scintillation crystals, time-of-flight and improved reconstruction algorithms resulting in substantial improvement in image quality. With experience, improved recognition of different patterns, such as the significance of focal versus heterogeneous or homogeneously increased marrow activity within particular lymphoma subtypes, has also increased accuracy. Earlier studies need to be reconsidered in light of these multiple advances.

In the current issue, Richardson et al. assessed the role of BMB in the FDG PET/CT era in a population of 50 patients with classical HL and also assessed current practice by surveying practitioners in 23 centers in the United Kingdom [14]. The cohort had a high risk of BMI, with an incidence of 20% compared to an overall incidence of 4–10% generally seen in HL [15]. All BMB-positive cases were identified by FDG PET/CT, and BMB did not identify involvement in any patient with normal marrow FDG uptake, suggesting the redundancy of BMB in patients staged with PET. Although PET identified patients with marrow involvement not seen with BMB, in this study, these findings did not result in treatment intensification, as these patients already had stage IV disease on the basis of CT staging. In their survey of practice, there was near universal (97%) routine use of BMB in advanced stage HL, contrasted by the low management impact demonstrated in the cohort studied. There was also a relatively high use of BMB in patients with limited HL, but this was significantly lower in the 50% of practitioners who used FDG PET/CT routinely.

Richardson's data are supported by several other recent studies indicating that in patients with classical HL, BMB can be safely omitted if PET/CT does not demonstrate marrow involvement [16–18]. There are particularly strong data in the pediatric population, where the incidence of marrow involvement is lower [19]. It is important to recognize that a pattern of diffuse homogeneous increased marrow activity in HL is more likely reactive due to reactive or inflammatory change driven by cytokine release [18,20]. Recognizing this, the specificity of PET/CT is very high, with rare false-positives in patients with focal abnormalities, as the CT component can precisely localize the abnormality to marrow and identify other non-lymphomatous etiology such as fracture, Paget's disease or fibrous dysplasia. Importantly, there is also very high reporter agreement for the identification of marrow involvement with PET, attributable to the high lesion-to-background contrast [21]. Thus, routine biopsy confirmation of multifocal FDG-avid marrow abnormality is unwarranted, especially if the intensity of metabolic abnormality within marrow mirrors the intensity at other sites of known disease. If there is uncertainty, a pragmatic approach is to perform an early FDG restaging study, with response or progression supporting the initial finding of marrow involvement with high certainty.

There are less robust data on the value of FDG PET/CT compared to BMB in high grade NHL. Diffuse large B-cell lymphoma (DLBCL), the most prevalent subtype, involves marrow in 10–30% of cases and can be focal or diffuse in distribution [15]. In a prospective cohort of 110 patients with DLBCL, FDG PET demonstrated marrow abnormality in only 10 of the 21 patients with BMB positivity [19]. However, most of the PET-positive patients had concordant focal large transformed lymphoid cells, whilst the ‘false-negative’ PET studies were predominantly discordant diffuse small-cell lymphoid infiltrates, the latter clonally distinct and possibly representing background low grade disease. Thus, the negativity of PET was not simply a lack of sensitivity, but illustrative of its ability to characterize the type of marrow involvement. A recent study has confirmed that concordant but not discordant involvement on BM biopsy is a risk factor for poor outcome independent of the international prognostic index [22]. Further research is needed to assess the prognostic significance of marrow involvement detected by FDG PET/CT.

Low grade lymphomas such as follicular lymphoma (FL) have a higher likelihood of BMI compared to high grade NHL or HL [15], but conversely the prevalence of marrow involvement detected with FDG PET is lower [23]. Normal marrow demonstrates metabolic activity, which can be similar in intensity to that of involved sites of low grade lymphoma, thus masking any focal or diffuse abnormality. PET is unable to detect small volume microscopic disease below its spatial resolution, also contributing to the low sensitivity of PET in patients with diffuse marrow involvement. In some patients with low-grade lymphoma but moderate-to-high metabolic abnormality such as grade III follicular lymphoma, PET may have diagnostic and prognostic utility beyond BMB.

More important than identifying the presence of marrow involvement is whether it represents an independent adverse prognostic marker which can be utilized to better direct personalized care. It is increasingly clear that in some lymphoma subtypes such as classical HL, BMB is redundant in the PET era, signifying another important step in de-escalation of management in these patients with good outcomes [14]. In other lymphoma subtypes, the prognostic power of PET in relation to bone marrow involvement is evolving. PET provides a unique opportunity to further characterize marrow involvement by quantifying its distribution, extent and metabolic intensity. Further research is needed within each lymphoma subtype to assess whether this can be utilized to better direct treatment and improve outcomes.

Potential conflict of interest:

Disclosure forms provided by the authors are available with the full text of this article at www.informahealthcare.com/lal.