Early VEGF testing in inflammatory neuropathy avoids POEMS syndrome misdiagnosis and associated costs

Background Prompt diagnosis and early treatment prevents disability in Polyneuropathy Organomegaly Endocrinopathy Monoclonal-protein and Skin Changes (POEMS) syndrome. Delay in diagnosis is common with 55% of patients initially incorrectly diagnosed with chronic inflammatory demyelinating polyradiculoneuropathy (CIDP). Patients are often treated with intravenous immunoglobulin which is both expensive and ineffective in the treatment of POEMS. Testing patients with acquired demyelinating neuropathy with serum vascular endothelial growth factor (VEGF) more accurately identifies POEMS syndrome than the current standard of care. Incorporating VEGF testing into screening could prevent misdiagnosis and reduce costs. Methods We used observed treatment information for patients in the University College London Hospital’s POEMS syndrome database (n=100) and from the National Immunoglobulin Database to estimate costs associated with incorrect CIDP diagnoses across our cohort. We conducted a model-based cost-effectiveness analysis to compare the current diagnostic algorithm with an alternative which includes VEGF testing for all patients with an acquired demyelinating neuropathy. Results Treatment associated with an incorrect CIDP diagnosis led to total wasted healthcare expenditures of between £808 550 and £1 111 756 across our cohort, with an average cost-per-POEMS-patient misdiagnosed of £14 701 to £20 214. Introducing mandatory VEGF testing for patients with acquired demyelinating neuropathy would lead to annual cost-savings of £107 398 for the National Health Service and could prevent misdiagnosis in 16 cases per annum. Conclusions Misdiagnosis in POEMS syndrome results in diagnostic delay, disease progression and significant healthcare costs. Introducing mandatory VEGF testing for patients with acquired demyelinating neuropathy is a cost-effective strategy allowing for early POEMS diagnosis and potentially enabling prompt disease-directed therapy.


INTRODUCTION
Polyneuropathy Organomegaly Endocrinopathy Monoclonal-protein and Skin Changes (POEMS) syndrome is a rare but treatable cause of acquired peripheral neuropathy. Patients present with length-dependent sensorimotor neuropathy, with mixed axonal and demyelinating features on neurophysiology. [1][2][3][4] Fifty-five per cent of patients with confirmed POEMS are initially misdiagnosed as having chronic inflammatory demyelinating polyradiculoneuropathy (CIDP) and are treated with immunomodulatory therapies including steroids, intravenous immunoglobulin (IVIG) and plasma exchange (PLEX). 5 IVIG is ineffective and costly (approximately £42.50 per gram in 2020), 6 often requires day case or hospital inpatient stays and can result in minor or severe complications. Diagnostic complexity results in a median time to POEMS diagnosis of 14 months, by which time over 30% of patients require a wheelchair or are bedbound. 5 POEMS syndrome diagnosis relies on identification of a lambda light chain restricted paraprotein in combination with the typical neuropathy as hallmarks of disease. 7 Routine investigations to discover a monoclonal protein involve a serum protein electrophoresis (SPEP) and immunofixation. We have demonstrated in our UK cohort of 100 patients that the SPEP was positive in 55% of cases, and immunofixation in 78%. 5 It is common practice for laboratories to perform immunofixation only if a paraprotein is present on SPEP, despite studies indicating the superior sensitivity of immunofixation in detecting low-level monoclonal bands missed by conventional electrophoresis techniques. 8 9 Although modern high-resolution electrophoresis can be as sensitive as immunofixation, 10 it is not in widespread use and has not been tested in POEMS syndrome cases which classically manifest small but significantly relevant monoclonal gammopathies. The data from our clinical cohort demonstrate that a critical low-level monoclonal band would not have been detected in 23% of cases by SPEP methodology only. The disparity in SPEP and immunofixation techniques across laboratories, combined with differences in levels of sensitivity and results in such tests often being difficult to interpret and rely on.
Serum vascular endothelial growth factor (VEGF) of >1000 pg/mL has high sensitivity (100%) and specificity (93%) in the diagnosis of POEMS syndrome, particularly when a demyelinating neuropathy and lambda light chain paraprotein present together. Levels are often very high (median pre-treatment VEGF levels in our cohort was 3 594 pg/mL), 5 and although iron deficiency anaemia, infection or chronic hypoxic states raise VEGF, 11-16 very high levels found with a demyelinating neuropathy and lambda light chain are diagnostic. This room temperature stable serum test can be sent to specialist laboratories for measurement, costing approximately £50 per sample. 17

Neuromuscular
We argue that an immunofixation and VEGF should be part of routine testing for patients presenting with an acquired peripheral neuropathy and with slowed conduction velocities on nerve conduction studies, particularly in those with suspected CIDP.
This study aims to add to the evidence base supporting a change in the polyneuropathy diagnostic process to include VEGF, uniquely from a cost perspective. In particular, the study will estimate: I.
The cost of misdiagnosing POEMS syndrome patients with CIDP II. The incremental cost-effectiveness ratio (ICER) of a new POEMS diagnostic pathway.

METHODS
Our sequential cohort (n=100) was taken from the POEMS syndrome database of University College London Hospital (UCLH) which includes clinical, diagnostic and treatment data. We collected additional data on IVIG treatment from the National Demand Management Programme for Immunoglobulin database. 18

Costing analysis
The costing analysis focused on comparing the cost of patients directly diagnosed with POEMS syndrome, compared with patients diagnosed with POEMS subsequent to an incorrect CIDP diagnosis. For each activity leading up to a confirmed POEMS syndrome diagnosis (figure 1), we estimated the quantity of resources used and multiplied these by their respective unit costs. A list of all costing inputs used is included in online supplemental material I which presents all inputs and assumptions used in this analysis.
As detailed IVIG treatment data, including number of treatments and IVIG quantity prescribed, were only available for a subset of patients (n=26), we used information from the National Immunoglobulin Database to estimate the average cost of IVIG treatment per patient, and combined this with unit costs of PLEX and corticosteroids to estimate the total costs associated with an incorrect CIDP diagnosis for each patient. By multiplying average cost-per-POEMS syndrome patient misdiagnosed with CIDP by the number of misdiagnosed patients, we estimated the total cost associated with CIDP misdiagnoses across our cohort, following the National Institute for Health and Care Excellence guidelines. This is the

Current POEMS diagnostic pathway, including steps associated with a previous incorrect CIDP diagnosis
GP refers patient to neurology for specialist opinion. Neurologist requests neuropathy screening bloods and outpatient neurophysiological assessment.

Serum Protein Electrophoresis (SPEP) performed by lab technician.
Immunofixation performed by lab technician.
Patient presents for outpatient neurophysiological assessment performed by skilled operator (consultant, registrar or neurophysiology technician). Reported by consultant neurophysiologist.
Patient presents to the GP with neuropathy. GP performs routine blood tests and arranges follow-up appointment to review results and clinical progression.

If +ve
Neurologist reviews results and diagnoses patient Patient attends outpatient clinic for 1 -5 courses of IVIG treatment and/or plasma exchange, and/or receives a course of corticosteroids.
Neurology consultant reviews patient progress and prescribes further CIDP treatment or reconsiders diagnosis. Blood tested at UCLH using VEGF by lab technician costs of the excess activities indicated in the pink shaded box in figure 1.

Cost-effectiveness analysis
We used decision analytical modelling to compare the costeffectiveness of the current diagnostic algorithm when investigating a patient with acquired demyelinating neuropathy (standard of care, SOC) with an alternative diagnostic algorithm which includes VEGF testing and mandatory immunofixation as follows, and detailed in figure 1: 1. Current SOC: SPEP; if positive, immunofixation. 2. Intervention: SPEP and immunofixation. VEGF testing in electrophysiologically confirmed acquired demyelinating polyneuropathy.
We modelled an incidence cohort of 3635 patients with an inflammatory polyneuropathy as the study population, which we estimated to approximate the annual number of patients referred by a general practitioner to neuromuscular clinics with any inflammatory polyneuropathy in the UK. 16 This estimate was based on the demographically similar population and healthcare system of the Netherlands as the nearest available to the UK (online supplemental material IV). Patients transitioned through the decision tree according to test accuracy, misdiagnosis and treatment rates (online supplemental material II). We used a time horizon from presentation with polyneuropathy symptoms, until a confirmed, correct diagnosis. POEMS syndrome diagnosis is typically between 6 months and 2 years, and no discounting was applied. Input data and sources are described in online supplemental material III.
Our model estimated the cost associated with each diagnostic pathway, and number of patients with POEMS syndrome with a correct initial diagnosis. To evaluate cost-effectiveness, we calculated the ICER to give the added cost per additional correct POEMS syndrome diagnosis. We carried out one-way deterministic sensitivity analyses and probabilistic sensitivity analysis, as detailed in the online supplemental material III.

ETHICS
This study was approved by the London School of Hygiene and Tropical Medicine Ethics Committee. The retrospective cohort data of which this project became a part was approved by the Health Research Authority and London Queen Square Research Ethics Committee.
If we assume patients misdiagnosed with CIDP (n=55) with missing treatment information (n=15) received no treatment, the total costs of CIDP misdiagnosis across our cohort is £808 550 (average cost £14 701). However, if we assume these patients received treatment in the same proportions as the cohort for which treatment information is known (n=40), the total costs of CIDP misdiagnoses is £1 111 756 with a median cost of £20 214 per patient (IQR: £11 808-£30 348). (Calculated by multiplying recorded IVIG treatment data (grams per course, number of IVIG courses) by unit costs (online supplemental material 1) for each patient, and summing all estimates (£676 2431) and dividing by n (=26).)

Cost-effectiveness analysis
The intervention diagnostic algorithm, in which all patients with acquired demyelinating polyneuropathy were screened with VEGF (including SPEP and immunofixation), would save £107 398 and result in 15.6 additional patients with POEMS syndrome directly diagnosed per year across the UK (table 1). The sensitivity analysis shows that the intervention dominated the SOC across uncertainty values (online supplemental material VI).

DISCUSSION
Our study found that from a cohort of 100 patients with POEMS syndrome, 55 were initially diagnosed with CIDP. Treatment associated with an initial incorrect CIDP diagnosis led to large, wasted healthcare expenditure. Treatment with IVIG alone was estimated to cost £20 984 per POEMS syndrome patient incorrectly diagnosed with CIDP, and we estimated that between 69% and 95% of misdiagnosed patients received IVIG treatment. When combined with the PLEX and corticosteroid treatment costs for patients recorded to have received these, the total wasted healthcare expenditure of CIDP misdiagnoses across our 100-patient cohort was between £808 550 and £1 111 756. This is a substantial waste of resources, and given the National Health Service (NHS) is extremely resource constrained, carries a large opportunity cost. Incorrect IVIG treatment for patients with POEMS syndrome may also have resulted in unnecessary harmful side effects. Indirect costs, such as time lost at employment or education, travel costs for treatment, and the emotional and social impacts of diagnostic uncertainty were not evaluated in this study and thus the true cost to misdiagnosis is likely to be far higher than that calculated here.
Our cost-effectiveness analysis suggests misdiagnosis and associated costs could be reduced or avoided by a change in the diagnostic protocol. Introducing mandatory immunofixation with an SPEP for patients presenting with an inflammatory polyneuropathy, and VEGF testing for patients with an acquired demyelinating polyneuropathy (most often considered to represent CIDP) could immediately lead to annual cost savings of £107 398 for the NHS. This pathway would require an increased number of VEGF and immunofixation tests but would result in a higher number of patients with POEMS syndrome initially correctly diagnosed and therefore reduced waste expenditure for the treatment of incorrect conditions.

RECOMMENDATIONS
Routine inflammatory neuropathy screening with SPEP only is not adequately sensitive to detect small plasma cell clones. Monoclonal gammopathies that are correctly identified are additionally at risk of misinterpretation as a paraproteinaemic neuropathy or coincidental monoclonal gammopathy of Neuromuscular unknown significance and thus IVIG treatment remains indicated. This study highlights the clinical and economic rationale first to test immunofixation in combination with SPEP in all cases presenting with inflammatory neuropathy. This is the most sensitive measure to identify relevant monoclonal gammopathies which may be associated with the neuropathy and require specific treatment. Once neurophysiology is performed, all cases of acquired demyelinating peripheral neuropathy (in which most are considered to be CIDP in the outpatients setting), particularly those where IVIG is being considered, should receive a VEGF test (see figure 1). A significantly raised VEGF at this stage would be a strong indication of POEMS syndrome and should prompt thorough exploration for an underlying monoclonal plasma cell disorder if not already discovered on initial serological testing. Mildly elevated VEGF can occur rarely in other inflammatory neuropathies and haematological malignancies, 11 and therefore the combination of demyelinating neuropathy, significantly raised VEGF and lambda paraprotein is essential to make a definitive POEMS diagnosis. Patients in our retrospective cohort diagnosed in less than 6 months from symptom onset had significantly lower OverallNeuropathy Limitation Scores (n=4) compared with those diagnosed after 6 months (n=6) (p<0.05) suggesting delayed diagnosis increases neuropathy severity. 5 Implementation of VEGF testing into routine clinical practice should correctly identify more POEMS cases from CIDP on initial presentation and avoid ineffective immunomodulatory therapy. Early diagnosis will allow for initiation of POEMSdirected therapy resulting in improved patient outcomes. The ultimate objective of this newly proposed management strategy is to improve patients' quality of life, and ability to live and work independently.
Contributors ESM collected data, performed the primary analysis and wrote the manuscript. SK was responsible for the paper concept and design, data collection, writing the manuscript and edits. FT-P performed data analysis and edits. SD was responsible for data acquisition and edits. MPL was responsible for the paper concept and design, data interpretation and edits.