Diagnosis of de novo HL was made by biopsy of sites of primary lymph node or extranodal disease. Under 18-year-old and pregnant patients were excluded. All patients were staged by unilateral BMB and by FDG-PET/CT scans. Bone marrow from the iliac crest was biopsied and assessed by a local senior hematopathologist.

Whole-body FDG-PET/CT imaging was acquired following standard protocols regarding uptake time (60–90min) after the intravenous administration of 296–444 MBq (8–12mCi) of FDG with a maximum interval of 14 days between BMB and PET scanning. Scanners were calibrated and images scaled for reading according to local protocols. Staging scans were performed prior to treatment.

FDG-PET/CT was classified as negative for bone marrow involvement in the absence of any focal area of increased bone marrow uptake or in the presence of diffuse bone marrow uptake. FDG-PET/CT was considered positive in the presence of focal FDG uptake regardless of diffuse uptake.

PET scans were centrally reviewed by two nuclear medicine physicians (JJC, CACPN). For the purpose of this study, all scans having abnormal FDG-uptake in bone marrow recorded in the database were examined to confirm and classify the pattern of marrow involvement.

Since histological examination of all sites is not possible, FDG-PET/CT and BMB results were combined in a reference standard in order to establish the diagnostic accuracy of the methods. Positive concordant findings at FDG-PET and BMB were interpreted as true positives. Concordant negative findings were interpreted as true absence of disease. In discordant cases with focal FDG uptake in marrow, resolution of the abnormal uptake in response to chemotherapy was taken as evidence of true marrow disease, even with no evidence of bone marrow involvement by histological analysis (biopsy findings in these cases were considered not representative, i.e., false negatives). FDG-PET with no apparent marrow disease was considered false negative in patients with a positive BMB.

The diagnostic accuracies were analyzed in terms of sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV). The Statistical Package for the Social Sciences (SPSS) version 10.0 for Windows (SPSS Inc., Chicago, IL) was used for the statistical analysis.

Based on information from both PET and BMB, 24% of cases were considered to have bone marrow involvement by HL. As reported by others, PET had greater sensitivity than BMB (PET 98% vs. BMB 40%) with the same specificity (100%). PET identified marrow disease not identified by BMB in 37/62 (59.6%) cases, which translated into upstaging patients to stage IV in 9% of the cohort.

In a metaanalysis involving five studies regarding PET bone marrow assessments in HL patients (with a total of 191 patients), Pakos et al.21 reached independent estimates for sensitivity and specificity of 76% and 92%, respectively. In a similar, more recent metaanalysis, Adams et al.7 reported pooled estimates of 96.9% and 99.7%, respectively. The results of this study were quite similar with sensitivity and specificity values for bone marrow disease detection by PET of 95% and 100%, respectively. Furthermore, concordant results between BMB and FDG-PET have been reported in the range of 70–80%,21,22 also similar to the findings of this study in Brazil [206/246 (83%) concordant PET and marrow biopsy results]. It is interesting to point out the similar high specificity of FDG-PET in this study as in the mentioned metaanalyses although the prevalence of infectious diseases potentially affecting bone marrow is higher in Brazil than in developed countries. This can be explained by the low incidence of involvement of the bone marrow by tropical infectious diseases; reports of simultaneous lymphoma and tropical infections are anecdotal.11,20 In addition, bone marrow involvement by these infections are more common in immunocompromised individuals, such as those co-infected with human immunodeficiency virus (HIV) or with allograft transplants16; these patients were not included in the current study.

The 1989 Cotswold modification of the Ann Arbor staging criteria abandoned the routine use of invasive procedures such as explorative laparotomy for HL staging, in the wake of improved imaging diagnosis with CT.23 Routine BMB was restricted to patients with CT-assessed stage III/IV disease or stage II disease with adverse ‘unfavorable’ factors, and only then if a positive finding would alter the therapeutic conduct. Although very few patients with early stage HL have a positive BMB24,25 some guidelines still recommend the inclusion of BMB in the routine staging work-up of patients with newly diagnosed HL.26 In the general practice, it is common to perform routine BMB for advanced stage HL due to the higher prevalence of bone marrow involvement, despite the fact that in advanced stage disease a positive BMB is much less likely to have a therapeutic impact than in early stage disease.27,28 The reason to perform BMB is the possible impact on treatment strategy. The reasons not to perform BMB are: (1) it is an unpleasant procedure for the patient, (2) it may delay the initiation of chemotherapy and (3) cost, especially in less developed countries. On the basis of these results BMB would have had an impact on the treatment of only a single patient if staging was performed with CT only, but with PET/CT staging, BMB had no therapeutic impact on any of the 246 patients.29 These results are also in line with a larger study presented recently by El-Galaly et al.13 These authors reviewed the experience of three Danish institutions where both BMB and PET/CT have been performed routinely to stage all patients with HL for several years. The study included 392 patients with HL, including 202 patients with early stage disease and 190 patients with advanced stage disease (according to PET/CT-based staging). Not a single patient with early stage disease had a positive BMB. Four patients with advanced stage disease had positive BMB despite normal FDG uptake in the bones and marrow. However, the treatment strategy was unaffected in all four patients. Therefore, similar to what was observed in this study, BMB was performed on several HL patients without a single therapeutic consequence.

In Brazil, PET was introduced later in comparison to developed countries, mainly due to financial reasons. Regarding the workup for HL, even after promising results from the first international PET studies, it was only after strenuous verification of its local cost-effectiveness that PET started to be incorporated into Brazilian practice.8 The present study corroborates the cost-effectiveness of FDG-PET in HL in Brazil. It demonstrates that, even in a tropical developing country with higher prevalences of advanced HL and infectious diseases, FDG-PET is accurate to identify clinically important bone marrow involvement by HL and can replace routine BMB in most cases. Staging BMB may be restricted to patients whose images or laboratory tests are suggestive of bone marrow involvement and in those with a potential change in treatment.