Lunn - Neuropathies and Paraproteins - AAM
Lunn - Neuropathies and Paraproteins - AAM
Lunn - Neuropathies and Paraproteins - AAM
Centre for Neuromuscular Disease, National Hospital for Neurology and Neurosurgery, Queen
michaellunn@nhs.net
Acknowledgements:
Dr Michael Lunn receives support from the NIHR UCL/UCLH Biomedical Research Centre.
Dr Michael Lunn has no financial conflicts of interest to declare in relation to this paper or its
contents.
Keywords:
The underlying pathogenesis of the haematological origin of the gammopathy causing anti-
MAG associated paraproteinaemic neuropathy may not be MGUS but more Waldenström’s
Macroglobulinaemia like opening possibilities for treatment advances.
Diagnostic and supportive criteria for POEMS are undergoing re-evaluations and although
excellent standards of diagnosis and treatment exist, recent advances in VEGF as a
biomarker and imaging findings are of great interest.
This is especially true in the peripheral nervous system where, to an extent, proof of a link is easier
to achieve. The most sensitive low level paraprotein detection methods should always be employed
where a paraprotein is suspected. Peripheral nerves can be damaged not only by the immunological
targeting of the myelin by the paraprotein, but by deposition (light chain amyloid and cryoglobulins)
or direct infiltration (neurolymphomatosis). This has resulted in other defined paraprotein related
disease pathogeneses.
Recent findings:
Our opportunities for treating these patients are greater not only through better recognition of
disease but also treatments introduced from haematological research. Beyond rituximab,
combination therapies, proteasome inhibition and novel biological treatments are being described in
haematological practice with early efficacy in neurology. Important developments here should be
Summary:
This review of the current literature focuses not only on the long term outcome studies in anti-MAG
Neuropathies as a complication of paraproteinaemias have been recognised for decades. For many
years these associations were rare, and attributed to myeloma where serum paraproteins were
easily identified by the only available test of agarose gel protein electrophoresis. As techniques for
detecting low level paraproteins increased in sensitivity [1], so our understanding of associations as
otherwise unexplained neuropathies [2], about four times more than would be expected from the
age matched paraprotein prevalence. The class distribution was heavily weighted in favour of IgM
paraproteins and once the HNK-1 carbohydrate epitope of myelin associated glycoprotein (MAG)
(also found on SGPG/SGLPG and some other peripheral nerve molecules) had been described as a
possible target, interest in proving links between neurology and haematology grew rapidly.
There are now a large number of neuropathies with a variety of pathogeneses associated with
paraproteins and paraproteinaemic disorders [3, 4]. These include the IgM associated demyelinating
well as some MMNCB, CANOMAD and a few neuropathies with other defined targets),
reviews are already available [3, 4] and here I will discuss the most recent evidence about the direct
paraprotein disorders using thalidomide, bortezomib and other chemotherapeutic agents will not be
covered.
[5]. Many of the modified Witebsky criteria demonstrating pathogenicity of anti-MAG antibodies are
fulfilled [6].
The haematology behind the IgM paraprotein in anti-MAG neuropathy has been debated over many
years. Anti-MAG antibodies are part of intrinsic repertoire antibody production with sequences
possibly matured from germline [7]. Expansion of IgM anti-MAG clones is usually considered part of
regulation gene are associated with Waldenström’s Macroglobulinaemia [8]. In patients with IgM
anti-MAG neuropathy at least 60% of patients have the MYD88 mutation detected, and this is likely
to be an underestimate [9]. This compares to other IgM MGUS related diseases where MYD88 is
seldom found. The implication is that anti-MAG PDPN patients are more WM-like in their
neuropathies [10]. A superficially complex but useful algorithm for diagnosing neuropathies
associated with WM summarises the breadth of the neuropathies associated with IgM in WM
The underlying similarity or even commonality with WM is very likely to determine treatment in the
near future. Although two randomised controlled trials of rituximab for anti-MAG neuropathy
individually failed to show significant benefit [11, 12], in meta-analysis rituximab has demonstrable
efficacy [13]. Obinutuzumab, a fully humanised anti-CD20 antibody with altered fucosylation to
increase NK-cell activation is usually used with chlorambucil for treatment of CLL. Four patients have
recently been described as being treated with this agent [14, 15]). Two of these patients, who were
both treatment naïve and treated early in their disease, improved dramatically with concomitant
biochemical improvements. The other two had longer disease duration and possibly too much
axonal loss to improve. Novel anti-CD20 agents such as obinutuzumab in combination with
chlorambucil, now seldom used, may be worth revisiting and exploring further.
However other agents beyond, and in addition to, antiCD20 therapy including bendamustine, 1st, 2nd
and 3rd generation proteasome inhibitors (bortezomib, carfilzomib, ixazomib) and the BTK inhibitors
(ibrutinib), may all be useful in the future. Ueki et al (2018) and Gomez et al (2016) have recently
neuropathy [16, 17]. Neither of the patients treated were typical (one had an associated Bing-Neel
syndrome and one was an IgM lambda associated anti-MAG neuropathy whereas the usual
association is with an IgMκ), but both improved. Bendamustine is relatively toxic and thus
proteasome inhibitors aimed at the plasma cell population may be better tolerated. Bortezomib has
off target neuropathic effects, but Ixazomib (a 3rd generation proteasome inhibitor) in combination
with dexamethasone and rituximab has been shown to be highly effective in the treatment of WM in
a phase II study of 26 patients. This study only included 2 neuropathy patients but both improved or
stabilised and only mild neuropathic symptoms occurred in others as possible complications [18].
Ixazomib should be investigated for the treatment of neuropathy in the near future. What is
required for new trials of treatment is well phenotyped homogeneous cohorts and responsive
outcome measures. These are being developed in the IMAGiNe study currently underway [19].
Rituximab has now been used for over 15 years as a single agent in IgM PDPN, and data are
emerging on the long term outcomes of cohorts of treated patients [20-23]. Gazzola and colleagues
published their centre experience of treating 33 anti-MAG patients with rituximab. 2/3 (18) of these
were atypical in their presentation (with subacute or proximal weakness for instance) and these
atypical patients were preferentially treated from the whole department cohort. Only 1/3 of the
rituximab cohort responded to treatment and the responders tended to be those with proximal
weakness and subacute evolution. Gazzola’s cohort could not find support for previous findings of
response in patients earlier in disease evolution or serum anti-MAG titres or IgM level, but numbers
were small. Predicting treatment response to target therapy continues to be difficult. Campagnolo et
al retrospectively reviewed 33 patients from 5 centres, 2/3 of whom had MGUS and 1/3 WM.
Numbers were again small but about half of patients responded, some only to a second dose of
rituximab. Response was not correlated to the underlying IgM associated disease, but a lack of
response was accompanied by a failure to reduce the overall IgM level after treatment. Some
patients in the Campagnolo study required two rituximab treatments, as was found also in the
follow up to RiMAG [12, 23]. Iancu-Ferfoglia et al explored the long term (6-year) outcome of a
subset of 15 patients in the RiMAG study [12] from a single centre. Over the 6 years since RiMAG, of
the seven who had been treated with rituximab in the initial study only two required further
treatment, versus six of eight in the placebo treated group. The only clinical difference identifiable
between the groups in outcome measures was a significant worsening of the 10m walk time in the
placebo group. The lack of difference between the groups was attributed more to poorly responsive
outcome measures than a lack of actual difference. Overall it seems clear that rituximab is effective
in the short and long term in a significant proportion of patients, but that predicting or measuring
In 2018 the collaborative Francophone Anti-MAG Cohort Group published the largest series of
patients in the literature [22]. These patients were defined primarily by the presence of an anti-MAG
antibody defined by the Bühlmann ELISA. Interestingly, possibly as a result of the Bühlmann
sensitivity, nearly 16% did not have demyelinating electrical studies, and a number of patients in this
series had other neuropathic diagnoses including Guillain-Barré Syndrome, motor neurone disease
and hereditary neuropathy with liability to pressure palsies, all underlining the fact that anti-MAG
antibodies occur in the serum of many but do not necessarily result in a neuropathy. There was no
link to the Bühlmann titre in terms of disease features or response to treatment. Although there is a
broadening of neuropathy phenotypes associated with MAG antibodies, it remains unclear if the
antibodies identified by current ELISA technology are pathogenic in all the typical and atypical cases.
The potential identification of large numbers of patients without a pathological causative link
between antibodies and the neuropathy may be partially responsible for the response of only 1/3 of
patients to treatment?
Although relatively safe in single use, rituximab is not without potential adverse effects. Svahn et al
[22] reported that 11 of 92 rituximab treated patients worsened transiently after treatment, and
Doneddu et al [24] also reported a deterioration in tremor. These cases of adverse response join
numerous others in the older literature to remind us that increasing effective therapy can potentially
POEMS
Until the last 10 years Polyneuropathy, Organomegaly, Endocrinopathy, M-Protein and Skin Changes
(POEMS) syndrome was largely neglected as a condition too rare for consideration. Good reviews of
POEMS diagnosis and management have recently been published [25, 26]. The discovery 20 years
ago that Vascular Endothelial Growth Factor (VEGF) was strikingly raised in POEMS led to its
inclusion as a major diagnostic criterion [25]. Performing a serum VEGF (as well as a serum
immunofixation and urine Bence Jones Protein) in all patients with a demyelinating neuropathy
might improve POEMS pick up rates, speed diagnosis and reduce disability, by providing an easy
biomarker to distinguish POEMS from CIDP. A raised serum VEGF of >1000pg/ml, when found in
conjunction with an appropriate neuropathy and a lambda light chain paraprotein is 100% sensitive
and 92% specific to POEMS [27]. However a number of other pathologies can raise VEGF levels
including iron deficiency anaemia, bone marrow and bone involving neoplasms and hypoxic
disorders and thus these need to be taken into account to prevent overdiagnosis.
Another interesting distinguishing feature of POEMS is that of vascular leak. This is responsible for
peripheral oedema, ascites, pleural and alveolar fluid [28] and papilloedema. Fluid accumulation also
occurs in more unusual sites, which might possibly be of additional diagnostic help. Ziff et al.
29 (71%) had asymptomatic pachymeningeal thickening compared to 0/33 CIDP patients (Figure 2).
In four POEMS patients there were significant meningeal collections of fluid. Asymptomatic
pachymeningeal thickening may be useful as another supportive minor diagnostic finding assisting
differentiation of POEMS from its major differential of CIDP. In addition, this paper also points out
that 59% of POEMS patients has brachial or lumbar plexus thickening and signal change as did 35%
of the CIDP control patients. Thus utilising root and plexus imaging to support a diagnosis of CIDP, as
per the EFNS/PNS criteria [31], may well lead to an erroneous CIDP diagnosis and this should be
borne in mind.
The treatment of POEMS remains fairly straightforward, albeit confined to centres of excellence. An
excellent annual update review was published again in 2019 [25]. Localised disease is treated with
radiotherapy, whilst disseminated disease (>3 solitary lesions or bone marrow involvement) is
treated with systemic therapy, either melphalan-200 autologous stem cell transplantation (ASCT) or
a combination of lenalidomide, melphalan, bortezomib or thalidomide with steroids. The latter two
are frequently complicated by worsening neuropathy and thus avoided where possible.
AL Amyloid
AL Amyloid is a plasma cell proliferative disorder in which light chains deposit in insoluble β-pleated
sheets in many tissues. Its median age of presentation is 62 years and untreated survival is is about
24 months. However, the prognosis varies widely from a few months to many years depending upon
the severity of cardiac involvement at presentation and for those fit for stem cell transplantation the
median survival may exceed 10 years [32]. Light chain (AL) amyloid remains difficult to diagnose
especially when there is early and limited organ involvement; early symptoms may be non-specific
weight loss and fatigue. Amyloid should always be suspected in ‘CIDP’ associated with carpal tunnel
syndrome, cardiac arrhythmias, nephrotic level proteinuria or notable fatigue [32]. The light chain is
lambda in the great majority of cases and biopsy of tissue is required to make the diagnosis. Non-
invasive diagnosis may be possible with florbetapir or florbetapen-PET scans, more usually used to
identify amyloid in Alzheimer’s disease. These radiolabel tracers can identify amyloid deposits to
target biopsies to lymph nodes or nerve tissue where suspicion is high and a SAP scan is negative
[33, 34]. Diagnosis by targeted biopsy demonstrating typical apple-green birefringent congo red
Rare localised forms of AL amyloid exist, and if truly local may not affect life expectancy and may not
require treatment. When there is systemic AL amyloid and performance status allows, treatment is
aimed at destroying the light chain producing clone [35]. Melphalan ASCT is the preferred treatment
with best outcomes when patients are carefully selected. Thalidomide and bortezomib are found to
be too neurotoxic toxic, and carfilzomib cardiac toxicity limits use. Daratumumab produces
haematological responses in 80% of patients and its further use in neuropathic AL amyloid is
awaited.
Cryoglobulinaemic neuropathies
Cryoglobulins are only found when correct procedures are used to collect and transport blood to the
laboratory as some precipitate at only just under 370C. Blood should always be collected,
transported and pre-processed in the laboratory at 370C. Breaks in the warm chain precipitate and
discard the cryoglobulin. Cryoglobulins are classified into three types: isolated monoclonal
cryoglobulinaemias are composed of either monoclonal and polyclonal Igs (type II) or polyclonal Igs
(type III) and are associated with chronic infections (usually hepatitis C virus) and connective tissue
disorders [36].
In a 2019 series of 492 cryoglobulinaemias from the Cleveland Clinic, 55% had a neurological
diagnosis, and 27% of 492 (131) were considered to have a definite or probable cryoglobulinaemia
link to the neurology. 15% of these were associated with an MGUS, and 92% of the 131 had a
neuropathy. Interestingly only 13% were a mononeuritis multiplex phenotype and only 55% had
pain. Most were symmetrical large fibre or small fibre neuropathies and might have been missed
without a high level of suspicion of testing for a cryoglobulin [36]. 102 patients all with Type I
cryoglobulinaemia were collected in another series [37]. In this paper the additional clues to assist
with diagnosis are emphasised, any or all of which may require careful clinical searching. They
include the presence of purpura (42%), and skin ulceration or gangrene (34%) being common, but
related disease, which is to remove or reduce the source. Chemotherapy regimens are directed at
the paraproteinaemic disorder and therefore include steroids, alkylating agents, rituximab (often in
combinations), novel myeloma therapies or autologous stem cell transplant [37]. To prevent
immediate tissue damage, either apheresis or cryofiltration can be tried. Immediate plasmapheresis
was effective in 123 of 159 cases in a series from Marson et al.[38]. Cooling the extracorporeal
circuit to form a cryogel precipitate may be more effective still [39] but is still yet to be brought into
mainstream use.
Neurolymphomatosis
The direct involvement of peripheral nerves by a paraproteinaemic clone is rare. It is most common
with non-Hodgkin’s B-Cell lymphomas (NHL), frequently diffuse large B-cell lymphoma (DLBCL)
subtypes. Less than 0.5% of NHL present with neurolymphomatosis [4]. However a painful multiple
mononeuropathy in the context of active or previously treated lymphoma, or in the context of a new
paraprotein should always prompt suspicion. Histopathological confirmation of the diagnosis is the
gold standard, but successfully targeting the right peripheral nerve site requires great experience
and some luck. However imaging techniques are developing enabling better localisation. In 2010, the
International Primary CNS Lymphoma Working Group identified 77% of MRI and 84% of PET-CT were
abnormal in a series of 50 neurolymphomatosis patients [40]. More recently a single centre
collection of 25 patients provided detailed neurography findings supporting the diagnostic utility
combined PET-CT and MRI imaging. In proven cases the bone marrow biopsy was negative in 17/19
(89%) and CSF cytology negative in 16/23 (70%). Imaging finding were much more helpful and
included the presence of a linear or fusiform mass (23/26, 88%) on 18F-FDG PET/CT which was only
moderately avid (average SuVmax: 7.1 ± 4.5, range, 1.5-17.0). On MRI T2-weighted scans masses were
hyperintense (21/22, 95%), enhancing (21/22, 95%) and linear or fusiform (19/22, 86%), with
associated muscle denervation in a significant number (14/22, 64%) [41]. Therefore biopsy targeted
by imaging may well improve and speed diagnosis in cases where suspicion is high.
Conclusion
The links between haematologists and neurologists are increasingly frequent as the recognition of
paraprotein related disorders affecting peripheral nerves becomes more evident. Many are difficult
to diagnose, especially when the pathology is confined to the peripheral nerves alone. However
improved biochemical paraprotein detection, and better imaging and targeted biopsy where
required will benefit earlier and broader therapeutic opportunities developing in the haematology
[1] Dispenzieri A, Kyle R, Merlini G et al. International Myeloma Working Group guidelines for serum-
free light chain analysis in multiple myeloma and related disorders. Leukemia 2009; 23:215-224.
[2] Kelly JJ, Jr, Kyle RA, O'Brien PC, Dyck PJ. The prevalence of monoclonal protein in peripheral
A very good, detailed overview of the neuropathic complications of lymphoma with detailed
descriptions of phenotypes.
[5] Nobile-Orazio E, Meucci N, Baldini L et al. Long-term prognosis of neuropathy associated with
anti-MAG IgM M-proteins and its relationaship to immune therapies. Brain 2000; 123:710-717.
[6] Witebsky E, Rose NR, Terplan K et al. Journal of the American Medical Association 1957;
164:1439-1447.
[7] Spatz LA, Williams M, Brender B et al. DNA sequence analysis and comparison of the variable
heavy and light chain regions of two IgM, monoclonal, anti-myelin associated glycoprotein
121:4434-4436.
[9] Vos JM, Notermans NC, D'Sa S et al. High prevalence of the MYD88 L265P mutation in IgM anti-
1009.
*[10] D'Sa S, Kersten MJ, Castillo JJ et al. Investigation and management of IgM and Waldenstrom-
An international consensus on the diagnosis and management of neuropathy associated with WM.
Contains a decision tree for diagnosis of the commonest associated neuropathic syndromes.
[12] Leger JM, Viala K, Nicolas G et al. Placebo-controlled trial of rituximab in IgM anti-myelin-
5:e460.
[15] Briani C, Visentin A, Salvalaggio A et al. Obinutuzumab, a new anti-CD20 antibody, and
[16] Ueki S, Nakamura M, Sasaki R et al. Beneficial Effect of Bendamustine in a Patient with Anti-
Biclonal IgM kappa and IgM lambda: Case Report and Review of the Literature. Clin Lymphoma
**[18] Castillo JJ, Meid K, Gustine JN et al. Prospective Clinical Trial of Ixazomib, Dexamethasone,
and Rituximab as Primary Therapy in Waldenstrom Macroglobulinemia. Clin Cancer Res 2018;
24:3247-3252.
A crucial paper of a phase II trial of ixazomib, dexamathosone and rituximab in the treatment of
only 2 included patients had neuropathy. A further phase III trial is awaited in which hopefully
neuropathic outcomes will be collected to identify if there is a therapeutic benefit to this new agent.
[19] Pruppers MHJ, Merkies ISJ, Lunn MPT et al. 230th ENMC International Workshop:: Improving
future assessment and research in IgM anti-MAG peripheral neuropathy: A consensus collaborative
effort, Naarden, The Netherlands, 24-26 February 2017. Neuromuscul Disord 2017; 27:1065-1072.
[20] Campagnolo M, Zambello R, Nobile-Orazio E et al. IgM MGUS and Waldenstrom-associated anti-
MAG neuropathies display similar response to rituximab therapy. J Neurol Neurosurg Psychiatry
2017; 88:1094-1097.
[21] Gazzola S, Delmont E, Franques J et al. Predictive factors of efficacy of rituximab in patients with
**[22] Svahn J, Petiot P, Antoine JC et al. Anti-MAG antibodies in 202 patients: clinicopathological
The largest cohort of anti-MAG associated patients in the literature collected partly retrospectively
from 14 French centres. The report contains detailed data of 202 patients and their response to
treatment, found nowhere else in the literature. However it also serves as an example of a disease
cohort defined by an antibody, and as the antibody may not always be pathogenic, the clinical
[23] Iancu Ferfoglia R, Guimaraes-Costa R, Viala K et al. Long-term efficacy of rituximab in IgM anti-
myelin-associated glycoprotein neuropathy: RIMAG follow-up study. J Peripher Nerv Syst 2016;
21:10-14.
[24] Doneddu PE, Kazmi M, Samuel M et al. Deterioration of tremor after treatment with rituximab
Am J Hematol 2019.
An excellent summary in the state of the art in POEMS. This paper contains execllent reference
*[27] Pihan M, Keddie S, D'Sa S et al. Raised VEGF: High sensitivity and specificity in the diagnosis of
All diagnostic tests and biomarkers can be misused. This paper illustrates the utility of VEGF in the
diagnosis of POEMS considering all the appropriate differential diagnoses as well as the confounders
for finding a raised VEGF level. VEGF should be utilised with a search for an appropriate monoclonal
plasma cell disorder and interpreted in the light of that and the neuropathy as in the published
criteria.
[28] Al-Mayoof O, Al Sughaiyer H, Abuomar W, Khan M. POEMS syndrome: a rare cause of exudative
ascites and chronic peripheral neuropathy. BMJ Case Rep 2017; 2017.
[29] Briani C, Fedrigo M, Manara R et al. Pachymeningeal involvement in POEMS syndrome: MRI and
*[30] Ziff OJ, Hoskote C, Keddie S et al. Frequent central nervous system, pachymeningeal and plexus
It would not be usual to image the brain in a peripheral nerve disorder where there are no clinical
indicators to do so. However vascular leak occurs in the meninges as it does in other tissues and in
some respects is easier to identify there than at some other sites. Furthermore the presence of
signal change and enhancement in the plexi in POEMS, currently suggested as a diagnostic feature of
CIDP means that the current criteria could actively result in POEMS misdiagnosis as CIDP. This paper
has important implications for future POEMS and CIDP criteria sets.
[31] Van den Bergh PY, Hadden RD, Bouche P et al. European Federation of Neurological
Societies and the Peripheral Nerve Society - first revision. Eur J Neurol 2010; 17:356-363.
[32] Gertz MA. Immunoglobulin light chain amyloidosis diagnosis and treatment algorithm 2018.
*[33] Broski SM, Spinner RJ, Howe BM et al. 18F-Florbetapir and 18F-FDG PET/CT in Systemic
Immunoglobulin Light Chain Amyloidosis Involving the Peripheral Nerves. Clin Nucl Med 2016;
41:e115-117.
Novel imaging methods may result in an otherwise cryptic diagnosis being made and treatment
being offered much earlier. This repurposing of an amyloid PET tracer can be very effective.
[34] Fox TA, Lunn M, Wechalekar A et al. [(18)F]Florbetaben PET-CT confirms AL amyloidosis in a
[35] Vaxman I, Bernstine H, Kleinstern G et al. FDG PET/CT as a diagnostic and prognostic tool for the
An impressively large series of patients with cryoglobulinaemia diligently collected over many years.
A quarter of the patients have definite or probable neurology associated with the cryoglobulin
illustrating the significant pathogenic potential of precipitating protein. Most of the neurology is
unsurprisingly neuropathic, but given the pathology is vasculitic the lack of pain in most and the
majority being symmetrical sensorimotor neuropathy make this an important illustration of the need
[37] Sidana S, Rajkumar SV, Dispenzieri A et al. Clinical presentation and outcomes of patients with
multicentre cohort study of 159 patients. Transfus Apher Sci 2018; 57:639-645.
[39] Taniyama Y, Nakatani Y, Matsuoka T et al. Efficacy of Cryofiltration for Treatment of Mixed
**[41] DeVries AH, Howe BM, Spinner RJ, Broski SM. B-cell peripheral neurolymphomatosis: MRI and
The utility of an imaging modality is only as good as its resolution powers. Using both PET-CT and
MRI together, targeted to suspicious areas increases diagnostic yield and leads to a refinement of
diagnosis. In combination with more aggressive chemotherapy regimens that allow better
convexities. b Coronal post-gadolinium T1w equivalent plane image from the same patient
demonstrating thick enhancing dura in this area. c Post-gadolinium T1w coronal image showing two
opposing layers of enhancing dura. The central non-enhancing area is effusion: the “double falx”
pachymeninges. f Coronal FLAIR showing moderate degree of small vessel disease. g Post-
gadolinium T1w image of the cervical spine level C5 showing intradural leptomeningeal
enhancement of cervical nerve roots. h Thickened and enhancing roots/proximal lumbar plexus on