What this study adds: This study highlights how suboptimal diagnostic pathways delay tuberculosis and lymphoma diagnosis in HIV-endemic settings. It underscores the limited value of FNA cytology and underuse of GeneXpert on FNA.

South Africa is the epicentre of the dual tuberculosis and HIV epidemics. Patients presenting with persistent lymphadenopathy and constitutional symptoms in the public healthcare system are deemed, in the first place, to suffer from extra-pulmonary tuberculosis (EPTB), regardless of age or HIV status.¹ Less diagnostic consideration is given for malignant causes of lymphadenopathy, such as haematological malignancies, predominantly lymphoma, and solid cancer metastases to the lymph nodes.^2,3

In people with HIV (PWH), EPTB accounts for up to 50% of tuberculosis cases, with tuberculosis adenitis the most common site of infection.^1,3,4 The diagnosis of EPTB requires specific and sensitive tests to account for the pauci-bacillary nature of the disease. The introduction of the GeneXpert MTB/RIF assay (Xpert) enhanced diagnostic accuracy for active tuberculosis in lymph nodes and heralded an era of rapidly evolving role of molecular diagnostics. Subsequently, Xpert Ultra, and other proven tuberculosis nucleic acid amplification tests (TB-NAATs), have further improved test sensitivity, while point-of care testing further reduced time to diagnosis.^5,6,7 Since Xpert’s inclusion in the 2013 WHO guidelines, TB-NAATs have been recommended for first-line assessment of EPTB.⁸ Despite this strong recommendation, TB-NAATs have not been widely implemented in the investigation of lymphadenopathy in our region.^9,10 In common practice, lymphadenopathy is investigated using fine needle aspirate (FNA) sent for either cytology, tuberculosis microscopy, or both.¹¹ The unacceptably low diagnostic accuracy of these tests leads to the failed detection of not only tuberculosis but also of lymphoma.^12,13 With both conditions inaccurately ‘ruled out’, a presumptive diagnosis of EPTB often follows, with empiric tuberculosis therapy started without definitive proof of mycobacterial infection, and without clinical follow-up to ensure response.^1,3

It is imperative that the clinician should not only consider EPTB as a differential diagnosis in patients with lymphadenopathy, but should also consider other diagnoses, notably lymphoma and solid cancer metastases. Considering these serious conditions as top differential diagnoses ensures that the clinician appropriately chooses the best tests to arrive at the correct diagnosis. Fine needle aspirate for cytology (FNAC) is a good first-line test for cancer, but has unacceptably low sensitivity for lymhoma and EPTB, which are more common, and more treatable.^3,12 Antel et al. showed that the use of FNAC (especially repeated use) significantly delayed the diagnosis of lymphoma by increasing the interval from presentation to diagnosis, termed the healthcare practitioner interval.¹

We hypothesised that the choice of tests performed in the diagnostic pathway of lymphadenopathy has an impact on diagnostic delay and increases the healthcare practitioner interval in patients referred for lymph node excision biopsies. We further wanted to analyse the tests related to the investigation of lymphadenopathy before excision biopsy, including tuberculosis investigations and tests related to aspiration of the lymph node. To investigate this, we conducted a retrospective review of all patients admitted to the surgical biopsy clinic at Groote Schuur Hospital (GSH) in 2016. These patients underwent a diagnostic excision biopsy of the lymph node, with the focus on obtaining a histological diagnosis.

In 2016, 166 patients were evaluated in the surgery clinic for ‘lumps and bumps’ (Figure 1). Of these patients, 86 had histological confirmation of a lymph node excision. The majority of these patients (71%, n = 61/86) had no previous diagnosis to explain the lymphadenopathy and were included in the study (Table 1 and Table 2). The remaining 25 patients, classified as ‘Previous diagnosis’, were undergoing repeat assessment for a previously confirmed haematological or non-haematological malignancy or tuberculosis and have been excluded from further analysis.

We describe the laboratory investigations performed prior to excisional lymph node biopsy and how inappropriate test selection may contribute to diagnostic delay. Our retrospective analysis reveals that FNA was frequently performed but mostly utilised for cytology, a test with poor sensitivity for the main differential diagnosis of tuberculosis and lymphoma. The preferred rule-out test for tuberculosis on FNA, Xpert, was never performed. In this cohort, histological assessment of excisional biopsies showed that over 60% of patients had one of three notable diagnoses, namely: tuberculosis, lymphoma, and metastatic cancer. The first two conditions respond well to prompt treatment, but poorly to late treatment, underscoring the urgency of improving pathways to timely and accurate diagnosis of these lymphadenopathy disease differentials.

The gold standard histological diagnosis of lymphadenopathy is obtained through a surgical excision biopsy. In this cohort, it took a median of 55 days (IQR 22–106) from presentation with persistent lymphadenopathy to a diagnostic biopsy. According to the UK NICE guidelines, 28 days is an acceptable target from presentation to diagnosis when a patient presents with suspected haematological malignancy.^1,15 These guidelines further specify that, ideally, a biopsy should be performed within 2–3 weeks. The recommended diagnostic interval was not achieved in this cohort. This confirms our previous findings that healthcare practitioner interval contributed significantly to diagnostic delay in lymphoma, with a median delay of 48 days from first healthcare contact to diagnostic biopsy and repeat FNAC associated with delay.¹ A total of 10 patients had repeat sputum tests for Xpert, where an Xpert on FNA would have been more appropriate. Fine needle aspirate for cytology was repeatedly performed and seldom gave a rule-in diagnosis, while the poor sensitivity of FNA for either TB or lymphoma made a confident rule-out diagnosis impossible. Furthermore, the interval from initial investigation to diagnostic biopsy was also unacceptably long: 49 days (IQR 14–80) for sputum, and 30 days (IQR 6–86) for FNA. Performing a test that is sensitive and specific from the very beginning reduces healthcare utilisation and improves timely diagnosis.¹⁶

Tests to diagnose tuberculosis adenitis in this patient cohort before biopsy were poorly chosen. Approximately half of the patients underwent FNA cytology, and, in a smaller proportion, a smear for AFBs was requested. Both these tests lack sensitivity for tuberculosis diagnosis because of the paucibacillary setting of lymph node infection.^17,18 The use of Xpert was confined solely to sputum, and was repeated in one third of cases. This, despite the presence of readily accessible lymphadenopathy. This practice contradicts the 2013 WHO Policy Update on Xpert which clearly recommends Xpert testing of non-respiratory specimens, including lymph node tissue, as the first-line diagnostic test for EPTB.^8,19 In two diagnostic comparison studies of tuberculosis adenitis, the sensitivity of Xpert (49.3% and 60.1%) was significantly higher than that of smear microscopy (14.5% and 27.8%).^17,20 More recently, Antel et al. reported 100% specificity and 70% sensitivity for tuberculosis using the Xpert Ultra assay in 43 patients with tuberculosis adenitis.⁹ When patients present with both pulmonary symptoms and significant lymphadenopathy, concurrent Xpert tests (or other TB-NAAT) should be done on both lymph node and sputum. This aligns with recent guidelines that emphasise the ability of concurrent testing to improve diagnostic accuracy.⁷ Lymph node biopsy should be performed timeously, by either core-needle biopsy or surgical excision, if the Xpert is negative, or if the early response to tuberculosis therapy is poor.

Lymphoma accounted for 21% of diagnoses. The use of FNAC in lymphoma diagnosis, which was performed in more than half of the cases, is strongly discouraged because of low test sensitivity, unless auxiliary tests are available to enhance accuracy.^21,22 A meta-analysis of 42 studies (1989–2012) that reviewed the sensitivity of FNAC to detect lymphoma reported only 12% diagnostic accuracy for lymphoma.³ Antel et al. reported an even lower sensitivity, with 11% (n = 63/90) of FNACs being merely ‘suggestive’ of lymphoma, and requiring a next step of histological confirmation for final diagnosis.¹ We confirm this poor performance with FNAC being non-diagnostic in this cohort, and at best one patient with ‘atypical’ FNAC findings. The demographic distribution of lymphoma and EPTB in this cohort justifies the focus on both diagnoses, in the evaluation of lymphadenopathy in patients of any age and any HIV status.⁹

In contrast to the lack of utility of FNAC for lymphoma, the use of FNAC in the solid cancer diagnostic pathway is well established. In patients of an older age, as the importance of metastatic cancer rises, FNAC has an acceptable sensitivity for a first-line investigation.²³ This sensitivity was not evident in the 15% with solid cancer in our cohort, likely as they were a highly selected group.

This study highlights the importance of a structured pathway in managing lymphadenopathy to optimise resource use.²¹ Furthermore, this analysis provided the proof of concept for the implementation of the Rapid Access Diagnostic Lymphadenopathy Clinic (RADLAC) at GSH at the end of 2017 – a physician-led clinic centred on the provision of early access to point-of-care core biopsy sent away for both histological analysis and Xpert testing.^1,2,3 By following the evidence-based diagnostic pathway shown in Figure 2, the RADLAC clinic reduced the healthcare practitioner interval to 27 days.³

Our study has limitations. The sample size was small, and was biased through considering a selected cohort of patients in whom diagnosis was not obtained through first-line investigations. Retrospective collection of study data most likely led to an underestimation of the duration of the healthcare practitioner interval. Data were limited to evaluation of visit records, laboratory testing, and documented history given, to ascertain the first date of healthcare intervention, Lastly, all participants had an accessible peripheral lymph node; therefore, our findings may not be generalisable to patients with EPTB or lymphoma with central lymph nodes.

This retrospective analysis of patients with persistent lymphadenopathy demonstrates how poorly structured diagnostic pathways increase healthcare utilisation and contribute to diagnostic delay in urgently treatable conditions. The median diagnostic interval of 55 days, more than double the UK NICE target of 28 days, results from inappropriate test selection, reliance on FNAC, repeated testing without due consideration of differential diagnoses, and perceived or real barriers to referral for lymph node excision biopsy.

Focused investigations including Xpert and histological examination, are required for at least two key diagnoses in patients with lymphadenopathy, namely tuberculosis and lymphoma. Implementing improved clinical pathways for patients with lymphadenopathy in South Africa’s public healthcare system will enhance timely diagnosis of both tuberculosis and lymphoma, potentially leading to improved patient outcomes.