Lymphoma: Understanding the diagnostic challenges and improving outcomes in a TBand HIV-endemic area

Background and aims Diagnosing lymphoma can be challenging: even in the best-resourced settings, lymphoma diagnosis may be delayed owing to the insidious onset of symptoms and/or difficulties in obtaining lymph node biopsy for diagnosis. In TB-endemic countries the diagnostic challenges in lymphoma may be further compounded by symptoms overlapping with those of TB and by resource limitations that may impede obtaining a diagnostic biopsy. New lymphoma diagnostic modalities, including the use of nextgeneration sequencing, are evolving rapidly and informing both prognosis and therapy in the field of lymphoma. These modalities of testing are likely to mean that less tissue is needed for assessment, or even that diagnosis can be made from peripheral blood. In this thesis we identify and describe local barriers in the diagnosis of lymphoma; and describe methods used to decrease the time-to-diagnosis of lymphoma and to subtype the most common lymphoma, diffuse large B-cell lymphoma (DLBCL). Our four main aims have been to: (i) Describe the pathway to a diagnosis of lymphoma in South Africa, with emphasis on the examination of local barriers to diagnosis (including overlapping symptomatology with TB) by retrospectively studying a cohort of patients with lymphoma. (ii) Investigate the diagnostic utility of the newest TB diagnostic test (the Xpert MTB/RIF Ultra); and apply this test in a diagnostic algorithm for lymphadenopathy of unknown aetiology, both on a fine-needle aspirate (FNA) from a lymph node and on tissue obtained by corebiopsy. (iii) Investigate whether a rapid-access lymph node biopsy clinic that used a core-biopsy method for lymph node biopsy was able to reduce the time-to-diagnosis of lymphoma, and to subtype accurately lymphomas for which subtyping is clinically relevant. (iv) Describe the subtype of DLBCL by HIV status using an immunohistochemical algorithm, in order both to describe the pattern of DLBCL by the most current diagnostic classification and to lay the foundations for further genetic work potentially capable of identifying mutations that could be used for genetic testing in DLBCL locally. Methods Four cohorts were analysed: (i) In the first, a group of 163 patients with lymphoma, the time-to-diagnosis and factors causing a delay in diagnosis were analysed. (ii) In the second, 99 patients with lymphadenopathy of unknown cause were recruited and the sensitivity and specificity of the Ultra were analysed using both fine-needle aspirate (FNA) and a core-biopsy technique. (iii) In the third cohort (n = 130), which included the second cohort of n = 99), outcomes from a rapid-access lymph node biopsy clinic – which used the core-biopsy method for the diagnosis of lymphoma – were analysed. (iv) In the fourth cohort (n = 182), the tissue of patients with DLBCL was analysed based on further immunohistochemical stains using the Hans algorithm in order to determine the molecular subtypes of DLBCL and determine their effect on prognosis. Results (i) In the first cohort ((n = 163), 29% HIV-infected), which was studied retrospectively, it took a median of 7 weeks for the diagnosis of lymphoma to be made from the time the patient sought medical attention. The longest time delay was in the healthcare practitioner interval (time from first healthcare visit to diagnostic biopsy). On multivariable logistic regression analysis, diagnostic delay > 6 weeks was associated with late-stage disease at presentation (odds ratio (OR) 2.3, 95% confidence interval (CI) 1.1–5.2) and a diagnosis of Hodgkin lymphoma (OR 3.0, 95% CI 1.1–8.0). HIV status was not associated with diagnostic delay (OR 0.9, 95% CI 0.3–2.2). The median time to diagnosis for patients on TB treatment was a month longer than for patients not on TB treatment. However, this was not statistically significant owing likely to a small sample size (n = 16, p = 0.28). (ii) In the second cohort, the diagnostic utility of the Ultra on lymph node aspirate and tissue were prospectively evaluated. The Ultra was found to be a sensitive and specific test for diagnosing TB of the lymph node. When compared with culture, the Ultra on aspirate had a sensitivity of 78% (40–97; 7/9) and on tissue 90% (55–100; 9/10). When using a composite reference score that combined both ‘definite TB' (culture positive) and ‘probable TB' (histological and clinical criteria), the Ultra was superior to currently used methods of diagnosing TB of the lymph node, including the detection of AFBs from an aspirate or on tissue biopsy or tissue culture (which had sensitivities of 26%, 33% and 39% respectively). The detection of granulomas on histology had high sensitivity (83%) but the lowest specificity. (iii) In the third cohort, which was studied prospectively, we obtained two important outcomes. Firstly, we showed that a rapid-access lymph node biopsy clinic was able to decrease the time-to-diagnosis of lymphoma when compared with that of cohort one (the historical cohort) to a diagnostic interval of 13.5 days compared with 48 days (p = 0.002). Secondly, the core-biopsy was able to detect lymphoma with a high rate of accuracy. The first-attempt core-biopsy was able to diagnose lymphoma in 84% of cases, and provided sufficient tissue to subtype lymphoma in a high proportion of the lymphoma cases (27/30, 90%). (iv) In the fourth, retrospectively evaluated cohort of patients with DLBCL (n = 181, 51 HIVpositive), there was a similar distribution of germinal centre (GC) and activated B cell (ABC) subtypes in the HIV-infected and HIV-uninfected groups. In contrast to what is reported in HIV-negative DLBCL literature, there were no statistically significant differences in overall survival by DLBCL subtype. Moreover, no significant difference in 5-year overall survival was demonstrated between the GCB and ABC subtypes (HR 1.2, 95% CI 0.8–1.9). Patients with HIV infection with a CD4 count of < 150 CD4 cells/mm3 had significantly poorer survival than those with no HIV infection (HR 2.4, 95% CI 1.3–4.1). Conclusions Making a diagnosis of lymphoma in South Africa is challenging. Two of the greatest challenges are overlapping symptomatology with TB and obtaining an adequate lymph node sample. The Ultra test on lymph node tissue is rapid, sensitive and specific; and can be performed on both FNA and tissue obtained through core-biopsy. Making an accurate diagnosis of TB is of critical importance, as TB is the most common cause for enlarged lymph nodes in TB-endemic areas. A reliable TB test on lymph node tissue will enable both an accurate TB diagnosis and the identification of patients who are negative for TB. The latter will require further investigation with a lymph node biopsy (via core needle or excision). As diagnostic testing and methods evolve in the direction of next-generation platforms, it is important to understand the tumour biology in our local setting, and in HIV-lymphoma, in order to be able to apply these new methods. This thesis describes the subtypes and outcomes seen in HIV DLBCL – knowledge that will be important in developing genetic next-generation sequencing methods that are specific to the type of lymphoma to be found in our context (and which may include genetic mutations induced by the Epstein–Barr virus). A clear diagnostic algorithm for the investigation of lymphadenopathy is presented, pulling together the four aims presented at the start of this section. Considering the prevalence of TB in South Africa, the goal is, in the first instance, to diagnose or exclude TB adenitis. In patients who test negative for TB, the goal then becomes to proceed with a diagnostic biopsy and, from our findings, to support the use of core-biopsy. This method, we have demonstrated, is able to provide sufficient tissue for the diagnosis and subtyping of lymphoma