Developing a deep learning natural language processing algorithm for automated reporting of adverse drug reactions.

The detection of adverse drug reactions (ADRs) is critical to our understanding of the safety and risk-benefit profile of medications. With an incidence that has not changed over the last 30 years, ADRs are a significant source of patient morbidity, responsible for 5%-10% of acute care hospital admissions worldwide. Spontaneous reporting of ADRs has long been the standard method of reporting, however this approach is known to have high rates of under-reporting, a problem that limits pharmacovigilance efforts. Automated ADR reporting presents an alternative pathway to increase reporting rates, although this may be limited by over-reporting of other drug-related adverse events. We developed a deep learning natural language processing algorithm to identify ADRs in discharge summaries at a single academic hospital centre. Our model was developed in two stages: first, a pre-trained model (DeBERTa) was further pre-trained on 1.1 million unlabelled clinical documents; secondly, this model was fine-tuned to detect ADR mentions in a corpus of 861 annotated discharge summaries. This model was compared to a version without the pre-training step, and a previously published RoBERTa model pretrained on MIMIC III, which has demonstrated strong performance on other pharmacovigilance tasks. To ensure that our algorithm could differentiate ADRs from other drug-related adverse events, the annotated corpus was enriched for both validated ADR reports and confounding drug-related adverse events using. The final model demonstrated good performance with a ROC-AUC of 0.955 (95% CI 0.933 - 0.978) for the task of identifying discharge summaries containing ADR mentions, significantly outperforming the two comparator models.

Comparison of diurnal blood pressure and urine production between people with and without chronic spinal cord injury.

STUDY DESIGN: Observational study. OBJECTIVES: To quantify diurnal blood pressure (BP) patterns and nocturnal hypertension and to measure diurnal urine production in people with chronic spinal cord injury (SCI), compared with controls without SCI. SETTING: Chronic SCI population in the community in Victoria, Australia. METHODS: Participants were recruited by advertisement, and sustained SCI at least a year prior or were healthy able-bodied volunteers. Participants underwent ambulatory BP monitoring (ABPM), measurement of urine production, and completed questionnaires regarding orthostatic symptoms. Comparisons were made between participants with tetraplegia or paraplegia and able-bodied controls. Participants with night:day systolic BP < 90% were classified as dippers, 90-100% as nondippers, and >100% as reverse dippers. RESULTS: Groups with tetraplegia (n = 51) and paraplegia (n = 33) were older (42.1 ± 15 and 41.1 ± 15 vs. 32.4 ± 13 years, mean ± s.d.) and had a higher prevalence of males (88 and 85% vs. 60%) than controls (n = 52). The average BP was 110.8 ± 1.5/64.4 ± 1.2 mmHg, 119.4 ± 2.1/69.8 ± 1.5 mmHg, and 118.1 ± 1.4/69.8 ± 1.0 mmHg in tetraplegia, paraplegia, and controls, respectively. Of participants with tetraplegia, paraplegia and controls, reverse dipping was observed in 45, 13, and 2% (p < 0.001), while nocturnal hypertension was observed in 13, 23, and 18%, respectively (p = 0.48). A reduction in nocturnal urine flow rate compared with the day was observed in paraplegia and controls, but not tetraplegia. CONCLUSIONS: Similar to the effects of acute SCI, chronic SCI, specifically tetraplegia, also causes isolated nocturnal hypertension, reverse dipping, orthostatic intolerance, and nocturnal polyuria. Cardiovascular risk management and assessment of orthostatic symptoms should include ABPM.

Fosfomycin for Treatment of Prostatitis: New Tricks for Old Dogs.

Treatment options for prostatitis caused by multidrug-resistant gram-negative bacilli are limited. We report two cases cured with oral fosfomycin and provide a pharmacokinetic analysis of fosfomycin predose concentrations during treatment.

Optimal timing of oral fosfomycin administration for pre-prostate biopsy prophylaxis.

OBJECTIVES: As the optimal administration time for fosfomycin peri-procedural prophylaxis is unclear, we sought to determine optimal administration times for fosfomycin peri-procedural prophylaxis. METHODS: Plasma, peripheral zone and transition zone fosfomycin concentrations were obtained from 26 subjects undergoing transurethral resection of the prostate (TURP), following a single oral dose of 3 g of fosfomycin. Population pharmacokinetic modelling was completed with the Nonparametric Adaptive Grid (NPAG) algorithm (Pmetrics package for R), with a four-compartment model. Plasma and tissue concentrations were simulated during the first 24 h post-dose, comparing these with EUCAST susceptibility breakpoints for Escherichia coli, a common uropathogen. RESULTS: Non-compartmental-determined pharmacokinetic values in our population were similar to those reported in the package insert. Predicted plasma concentrations rapidly increased after the first hour, giving more than 90% population coverage for organisms with an MIC ≤4 mg/L over the first 12 h post-dose. Organisms with higher MICs fared much worse, with organisms at the EUCAST breakpoint being covered for <10% of the population at any time. Transitional zone prostate concentrations exceeded 4 mg/L for 90% of the population between hours 1 and 9. Peripheral zone prostate concentrations were much lower and only exceeded 4 mg/L for 70% of the population between hours 1 and 4. CONCLUSIONS: Until more precise plasma and tissue data are available, we recommend that fosfomycin prophylaxis be given 1-4 h prior to prostate biopsy. We do not recommend fosfomycin prophylaxis for subjects with known organisms with MICs >4 mg/L.

Network analysis of an in vitro model of androgen-resistance in prostate cancer.

BACKGROUND: The development of androgen resistance is a major limitation to androgen deprivation treatment in prostate cancer. We have developed an in vitro model of androgen-resistance to characterise molecular changes occurring as androgen resistance evolves over time. Our aim is to understand biological network profiles of transcriptomic changes occurring during the transition to androgen-resistance and to validate these changes between our in vitro model and clinical datasets (paired samples before and after androgen-deprivation therapy of patients with advanced prostate cancer). METHODS: We established an androgen-independent subline from LNCaP cells by prolonged exposure to androgen-deprivation. We examined phenotypic profiles and performed RNA-sequencing. The reads generated were compared to human clinical samples and were analysed using differential expression, pathway analysis and protein-protein interaction networks. RESULTS: After 24 weeks of androgen-deprivation, LNCaP cells had increased proliferative and invasive behaviour compared to parental LNCaP, and its growth was no longer responsive to androgen. We identified key genes and pathways that overlap between our cell line and clinical RNA sequencing datasets and analysed the overlapping protein-protein interaction network that shared the same pattern of behaviour in both datasets. Mechanisms bypassing androgen receptor signalling pathways are significantly enriched. Several steroid hormone receptors are differentially expressed in both datasets. In particular, the progesterone receptor is significantly differentially expressed and is part of the interaction network disrupted in both datasets. Other signalling pathways commonly altered in prostate cancer, MAPK and PI3K-Akt pathways, are significantly enriched in both datasets. CONCLUSIONS: The overlap between the human and cell-line differential expression profiles and protein networks was statistically significant showing that the cell-line model reproduces molecular patterns observed in clinical castrate resistant prostate cancer samples, making this cell line a useful tool in understanding castrate resistant prostate cancer. Pathway analysis revealed similar patterns of enriched pathways from differentially expressed genes of both human clinical and cell line datasets. Our analysis revealed several potential mechanisms and network interactions, including cooperative behaviours of other nuclear receptors, in particular the subfamily of steroid hormone receptors such as PGR and alteration to gene expression in both the MAPK and PI3K-Akt signalling pathways.

Positive surgical margins: rate, contributing factors and impact on further treatment: findings from the Prostate Cancer Registry.

OBJECTIVE: To describe the characteristics of patients with and without positive surgical margins (PSMs) and to analyse the impact of PSMs on secondary cancer treatment after radical prostatectomy (RP), with short-term follow-up. PATIENTS AND METHODS: We analysed data from 2385 consecutive patients treated using RP, who were notified to the Prostate Cancer Registry by 37 hospitals in Victoria, Australia between August 2008 and February 2012. Independent and multivariate models were constructed to predict the likelihood of PSMs. Independent and multivariate predictors of secondary treatment after RP in the initial 12 months after diagnosis were also assessed. RESULTS: Data on PSM status were collected for 2219/2385 (93%) patients. In total 592/2175 (27.2%) RPs resulted in PSMs; 102/534 (19.1%) in the low-risk group, 317/1218 (26.0%) in the intermediate-risk group, 153/387 (39.5%) in the high-risk group, and 9/11 (81.8%) in the very-high-risk disease group of patients. Patients having surgery in a hospital where <10 RPs occur each year were significantly more likely to have a PSM (incidence rate ratio [IRR] 1.44, 95% confidence interval [CI] 1.07-1.93) and those in the intermediate-, high- or very-high-risk groups (IRR 1.34, 95% CI 1.09-1.65, P = 0.007, IRR 1.96, 95% CI 1.57-2.45, P < 0.001 and IRR 3.81, 95% CI 2.60-5.60, P < 0.001, respectively) were significantly more likely to have a PSM than those in the low-risk group (IRR 2.50, 95% CI 1.23-5.11, P = 0.012). Patients with PSMs were significantly less likely to have been treated at a private hospital than a public hospital (IRR 0.76, 95% CI 0.63-0.93, P = 0.006) or to have undergone robot-assisted RP (IRR 0.69, 95% CI 0.55-0.87; P = 0.002) than open RP. Of the 2182 patients who underwent RP in the initial 12 months after diagnosis, 1987 (91.1%) received no subsequent treatment, 123 (5.6%) received radiotherapy, 47 (2.1%) received androgen deprivation therapy (ADT) and 23 (1.1%) received a combination of radiotherapy and ADT. Two patients (0.1%) received chemotherapy combined with another treatment. At a multivariate level, predictors of additional treatment after RP in the initial 12 months included having a PSM compared with a negative surgical margin (odds ratio [OR] 5.61, 95% CI 3.82-8.22, P < 0.001); pT3 compared with pT2 disease (OR 4.72, 95% CI 2.69-8.23, P < 0.001); and having high- or very-high-risk disease compared with low-risk disease (OR 4.36, 95% CI 2.24-8.50, P < 0.001 and OR 4.50, 95% CI 1.34-15.17, P = 0.015, respectively). Patient age, hospital location and hospital type were not associated with secondary treatment. Patients undergoing robot-assisted RP were significantly less likely to receive additional treatment than those receiving open RP (OR 0.59, 95% CI 0.39-0.88, P = 0.010). CONCLUSIONS: These data indicate an important association between hospital status and PSMs, with patients who underwent RP in private hospitals less likely than those in public hospitals to have a PSM. Patients treated in lower-volume hospitals were more likely to have a PSM and less likely to receive additional treatment after surgery in the initial 12 months, and robot-assisted RP was associated with fewer PSMs than was open RP in this non-randomized observational study. PSM status and pathological T3 disease are both important and independent predictors of secondary cancer treatment for patients undergoing RP. A robot-assisted RP approach appears to decrease the likelihood of subsequent treatment, when compared with the open approach.

The Prostate Cancer Registry: monitoring patterns and quality of care for men diagnosed with prostate cancer.

OBJECTIVE: To establish a pilot population-based clinical registry with the aim of monitoring the quality of care provided to men diagnosed with prostate cancer. PATIENTS AND METHODS: All men aged >18 years from the contributing hospitals in Victoria, Australia, who have a diagnosis of prostate cancer confirmed by histopathology report notified to the Victorian Cancer Registry are eligible for inclusion in the Prostate Cancer Registry (PCR). A literature review was undertaken aiming to identify existing quality indicators and source evidence-based guidelines from both Australia and internationally. RESULTS: A Steering Committee was established to determine the minimum dataset, select quality indicators to be reported back to clinicians, identify the most effective recruitment strategy, and provide a governance structure for data requests; collection, analysis and reporting of data; and managing outliers. A minimum dataset comprising 72 data items is collected by the PCR, enabling ten quality indicators to be collected and reported. Outcome measures are risk adjusted according to the established National Comprehensive Cancer Network and Cancer of the Prostate Risk Assessment Score (surgery only) risk stratification model. Recruitment to the PCR occurs concurrently with mandatory notification to the state-based Cancer Registry. The PCR adopts an opt-out consent process to maximize recruitment. The data collection approach is standardized, using a hybrid of data linkage and manual collection, and data collection forms are electronically scanned into the PCR. A data access policy and escalation policy for mortality outliers has been developed. CONCLUSIONS: The PCR provides potential for high-quality population-based data to be collected and managed within a clinician-led governance framework. This approach satisfies the requirement for health services to establish quality assessment, at the same time as providing clinically credible data to clinicians to drive practice improvement.

Patterns of care for men diagnosed with prostate cancer in Victoria from 2008 to 2011.

OBJECTIVE: To describe patterns of care for men diagnosed with prostate cancer in Victoria, Australia, between 2008 and 2011. DESIGN, SETTING AND PATIENTS: Men who were diagnosed with prostate cancer at 11 public and six private hospitals in Victoria from August 2008 to February 2011, and for whom prostate cancer notifications were received by the Prostate Cancer Registry. MAIN OUTCOME MEASURES: Characteristics of men diagnosed with prostate cancer; details of treatment provided within 12 months of diagnosis, according to National Comprehensive Cancer Network risk categories; and characteristics of men who did not receive active treatment within 12 months of diagnosis. RESULTS: Treatment details were collected for 98.1% of men who were assessed as eligible to participate in the study (2724/2776) and were confirmed by telephone 12 months after diagnosis for 74.4% of them (2027/2724). Most patients (2531/2724 [92.9%]) were diagnosed with clinically localised disease, of whom 1201 (47.5%) were at intermediate risk of disease progression. Within 12 months of diagnosis, 299 of the 736 patients (40.6%) who had been diagnosed as having disease that was at low risk of progression had received no active treatment, and 72 of 594 patients (12.1%) who had been diagnosed as having disease that was at high risk of progression had received no active treatment. Of those diagnosed as having intermediate risk of disease progression, 54.5% (655/1201) had undergone radical prostatectomy. Those who received no active treatment were more likely than those who received active treatment to be older (odds ratio [95% CI], 2.96 [2.01-4.38], 10.94 [6.96-17.21] and 32.76 [15.84-67.89], respectively, for age 65-74 2013s, 75-84 2013s and ≥ 85 2013s, compared with < 55 2013s), to have less advanced disease (odds ratio [95% CI], 0.20 [0.16-0.26], 0.09 [0.06-0.12] and 0.05 [0.02-0.90], respectively, for intermediate, high and very high-risk [locally advanced] or metastatic disease, compared with low-risk disease) and to have had their prostate cancer notified by a private hospital (odds ratio [95% CI], 1.35 [1.10-1.66], compared with public hospital). CONCLUSION: Our data reveal a considerable "stage migration" towards earlier diagnosis of prostate cancer in Victoria and a large increase in the use of radical prostatectomy among men with clinically localised disease.

Recent developments in prostate cancer biomarker research: therapeutic implications.

This review aims to present an overview of recent clinical trials targeting biomarkers in advanced prostate cancer. We searched ClinicalTrials.gov for early phase clinical trials on treatments of prostate cancer that have been recently completed, are ongoing or are actively recruiting participants. Drug targets and their mechanism of actions were assessed and summarized. Trials were categorized according to prostate cancer biomarkers that have potential as therapeutic targets. A total of 19 new therapeutic agents for the treatment of prostate cancer are included in this review. Trials are summarized according to the targeted biomarkers and are categorized into five therapeutic approaches: prostate cancer vaccine, epigenetic therapy, pro-apoptotic agents, prostate cancer antibodies and anti-angiogenesis approach. Some of the therapeutic agents reviewed showed promising results, warranting further investigation in late phase clinical trials. Recent novel prostate cancer biomarkers that made it through clinical trials and their relevance as drug targets are summarized. This review emphasizes the importance of specific prostate cancer biomarkers and their potentials as targets of the disease. Some clinical trials of targeted treatments in prostate cancer show promising results. Better understanding of disease mechanisms should potentially lead to more specific treatments for individual patients.

Gene sequencing and tissue expression of unknown isoforms of an angiotensin II type 2 receptor interacting protein, ATIP, in the rat.

To investigate the expression of the unknown angiotensin II type 2 receptor interacting protein (ATIP) isoforms in the rat we used the known sequences of human and mouse ATIP to design sequencing primers to enable us to sequence rat ATIP3 and ATIP4. Exon 4, which is present in human but not mouse ATIP, was not identified in the coding region of rat ATIP. The expression levels of these genes in a range of rat tissues were examined, and we concluded that there is little similarity in the relative tissue distribution of the various ATIP isoforms in rat and human.

Cure of Limb-Threatening XDR Pseudomonas aeruginosa Infection: Combining Genome Sequencing, Therapeutic Drug Level Monitoring, and Surgical Debridement.

We describe a case of limb-threatening osteomyelitis and metalware infection with carbapenemase-producing extensively drug-resistant Pseudomonas aeruginosa successfully cured with aggressive surgical debridement and combined intravenous fosfomycin and colistin. Real-time therapeutic drug monitoring was used to maximize probability of efficacy and minimize potential for toxicity.

Expression and function of ATIP/MTUS1 in human prostate cancer cell lines.

BACKGROUND: We have previously demonstrated Ang II type 2 (AT(2)-) receptor-mediated inhibition of EGF-induced prostate cancer cell growth in androgen-dependent (LNCaP) and independent (PC3) prostate cancer cell lines. METHODS: To explore the signaling pathways involved in this inhibitory effect, we examined the interaction of the AT(2)-receptor with its novel regulatory partner ATIP using real time PCR, over-expression, siRNA and [(3)H]thymidine incorporation assays. RESULTS: The results in human prostate cancer cell lines demonstrate the presence of ATIP in both cell lines examined, and suggest that (i) the AT(2)-receptor through an interaction with ATIP mediates an anti-growth factor effect in both androgen-dependent and androgen-independent cell lines; (ii) ATIP expression decreases as the rate of cell growth and androgen-independence increase; and (iii) EGF may act on cell growth in part by reducing the content of ATIP present in the cells. CONCLUSIONS: The results support our earlier proposal in normal cell lines that ATIP is an important component of the cellular response to AT(2)-receptor activation. The results further suggest that a critical level of ATIP is required to mediate the effect of AT(2)-receptor activation to inhibit EGF mediated increases in cell growth. They also suggest that EGF may in part induce cell growth by suppressing the level of ATIP expression.

A Machine-Learning Algorithm to Optimise Automated Adverse Drug Reaction Detection from Clinical Coding.

INTRODUCTION: Adverse drug reaction (ADR) detection in hospitals is heavily reliant on spontaneous reporting by clinical staff, with studies in the literature pointing to high rates of underreporting [1]. International Classification of Diseases, 10th Revision (ICD-10) codes have been used in epidemiological studies of ADRs and offer the potential for automated ADR detection systems. OBJECTIVE: The aim of this study was to develop an automated ADR detection system based on ICD-10 codes, using machine-learning algorithms to improve accuracy and efficiency. METHODS: For a 12-month period from December 2016 to November 2017, every inpatient episode receiving an ICD-10 code in the range Y40.0-Y59.9 (ADR code) was flagged for review as a potential ADR. Each flagged admission was assessed by an expert pharmacist and, if needed, reviewed at regular ADR committee meetings. For each report, a determination was made about ADR probability and severity. The dataset was randomly split into training and test sets. A machine-learning model using the random forest algorithm was developed on the training set to discriminate between true and false ADR reports. The model was then applied to the test set to assess accuracy using the area under the receiver operating characteristic (AUC). RESULTS: In the study period, 2917 Y40.0-Y59.9 codes were applied to admissions, resulting in 245 ADR reports after review. These 245 reports accounted for 44.5% of all ADR reporting in our hospital in the study period. A random forest model built on the training set was able to discriminate between true and false reports on the test set with an AUC of 0.803. CONCLUSIONS: Automated ADR detection using ICD-10 coding significantly improved ADR detection in the study period, with improved discrimination between true and false reports by applying a machine-learning model.

Correction to: A Machine-Learning Algorithm to Optimise Automated Adverse Drug Reaction Detection from Clinical Coding.

CK: Principal Investigator. CM and CK were responsible for the study design and conception; all authors were responsible for acquisition and validation of the data; CM was responsible for analysis and interpretation of the data; and all authors contributed to reviewing drafts of the manuscript and approved the final version for publication.

Association of good oncological response to therapy with the development of rheumatic immune-related adverse events following PD-1 inhibitor therapy.

AIM: To investigate whether any patient or treatment characteristics are associated with the development of rheumatic immune-related adverse events (irAEs) following programmed cell death protein 1 (PD-1) inhibitor therapy for cancer. METHOD: This was a retrospective chart review of all patients who were dispensed nivolumab or pembrolizumab at a single center before 1 January, 2017, with follow-up until 1 July, 2017. Patients with any diagnosis of a non-cutaneous irAE were identified, regardless of severity, and rheumatic irAEs were characterized. RESULTS: Of 244 episodes of therapy, a non-cutaneous irAE occurred in 72 (29.5%). Rheumatic irAEs were diagnosed in 19 episodes of therapy (7.8%), with 12 de novo diagnoses (5.1% of episodes without a pre-existing autoimmune rheumatic disease) and 7 exacerbations of existing disease. Review by a rheumatologist occurred in only 11 of these. Rheumatic irAEs were more common in patients with a good oncological response to therapy (relative risk [RR] 11.16), those being treated for melanoma (RR 2.94) and those who developed another non-cutaneous irAE (RR 2.64). CONCLUSION: Rheumatic irAEs are relatively common with PD-1 inhibitor therapy, and appear to be associated with a good oncological response to therapy. Many rheumatic irAEs were not referred to rheumatological services. Prospective systematic investigation would be of benefit to explore these characteristics.

Seeing the wood for the trees: approaches to teaching and assessing clinical pharmacology and therapeutics in a problem-based learning course.

For about 50 years, clinical pharmacology and therapeutics have been taught in the medical schools via traditional lectures and practical classes. During this time, significant changes have occurred in our understanding of medicine and basic sciences. Also the needs for our community have changed dramatically. The explosion of scientific discoveries, the use of new technologies in disease diagnosis, the availability of a wide range of therapeutic options, and the availability of knowledge to everyone via the Internet have necessitated new approaches for teaching medical and other health professional students. Finding information related to a topic has not become a priority in teaching, what has become more important is to teach undergraduate students how to think in addition to what to think. Applying information learnt and assessing its significance in real life situations has become mandatory. The aims of this paper were: (i) to discuss the model we used in introducing clinical pharmacology and therapeutics teaching in the undergraduate course at the University of Melbourne and the educational principles behind the model, and (ii) to discuss the new tools of assessment used in a problem-based learning (PBL) curriculum.

Rheumatic immune-related adverse events secondary to anti-programmed death-1 antibodies and preliminary analysis on the impact of corticosteroids on anti-tumour response: A case series.

IMPORTANCE: Rheumatic immune-related adverse events (irAEs) occur in approximately 10-20% of anti-programmed death 1 (anti-PD1)-treated cancer patients. There are limited data on the natural history, optimal treatment and long-term oncological outcomes of patients with rheumatic irAEs. OBJECTIVE: The objective of the study was to describe the spectrum and natural history of rheumatic irAEs and the potential impact of rheumatic irAEs and immunomodulators on anti-PD1 tumour efficacy. METHODS: Cancer patients with pre-existing rheumatic disease before anti-PD1 therapy or de novo rheumatic irAEs on anti-PD1 therapy were retrospectively reviewed across three sites. Patient demographics, treatment history, anti-PD1 irAEs, and anti-PD1 responses were evaluated. Relationships between the development or pre-existence of rheumatic irAE, use of immunomodulatory agents and outcomes were evaluated. RESULTS: This multicenter case series describes 36 cancer patients who had rheumatic disease before anti-PD1 therapy (n = 12) or developed de novo rheumatic irAEs (n = 24). Thirty-four of the 36 patients sustained rheumatic irAEs (median time to rheumatic irAE: 14.5 weeks), including 24 de novo (18 inflammatory arthritis, three myositis, two polymyalgia rheumatica, one fasciitis) and 10 flares in 12 patients with pre-existing rheumatic disease. Corticosteroids were used in 30 of 36 patients (median duration: 10 months), and disease-modifying antirheumatic drugs were used in 14 of 36 patients (median duration: 5.5 months). The objective response rate to anti-PD1 therapy was 69% (n = 25/36) overall and 81% (n = 21/26) in the melanoma subgroup. CONCLUSIONS: Rheumatic irAEs are often chronic and require prolonged immunomodulatory therapy. Prospective studies are required to define optimal management of rheumatic irAEs that maintain long-term anticancer outcomes.

Massive splenomegaly due to concurrent primary Epstein-Barr virus and cytomegalovirus infection in a patient on adalimumab.

A 32-year-old man who was receiving adalimumab for seronegative rheumatoid arthritis presented with a 4-week history of fever, night sweats, fatigue, myalgias and diarrhoea. On examination, he had obvious splenomegaly but no lymphadenopathy or pharyngitis. Full blood count revealed mild neutropenia and significant lymphocytosis, with a blood film showing atypical lymphocytes. Liver function tests were mildly deranged with a mixed hepatitic and obstructive pattern. Ultrasound confirmed massive splenomegaly with a span of 21 cm in the long axis. Serological tests confirmed the presence of both primary Epstein-Barr virus and cytomegalovirus infections. The patient had his adalimumab withheld, was treated with supportive measures and improved over a period of 8 weeks. He remained well 5 months after the onset of illness with complete normalisation of blood count and a resolution of the splenomegaly.