|Home | About | Journals | Submit | Contact Us | Français|
Paraneoplastic neurological disorders (PNDs) are a rare and diverse group of neurological conditions that can involve any part of the nervous system. Diagnosis is facilitated by finding well-recognized autoantibodies directed against neural antigens in the sera and the cerebrospinal fluid. Identifying and eliminating the underlying malignancy is the mainstay of treatment. Immunomodulatory treatment is gaining more acceptance especially, where a malignancy could not be identified, oncology treatment is completed, or along with cancer treatment. Literature review shows only a handful of systematic prospective case series. Multicenter, prospective controlled clinical trials are needed for future therapeutic advances.
Paraneoplastic neurological disorders (PNDs) are a group of heterogeneous neurological disorders in patients with cancer. These are due to remote immunological effects of malignancy and not related to metastases or direct invasion by the tumor. These neurological syndromes typically present before the diagnosis of cancer. Hence, early recognition is of paramount importance. Overall, symptomatic PNDs are rare, affecting perhaps 0.01% of cancer patients [Molina-Garrido et al. 2006; Darnell et al. 2003]. However, there are exceptions such as myasthenia gravis, which is seen in as many as 15% of patients with thymoma; Lambert-Eaton syndrome, which affects up to 3% of patients with small-cell lung carcinoma (SCLC) [Elrington et al. 1991]; and peripheral neuropathy, which affects half of patients with osteosclerotic plasmacytoma [Dispenzieri and Gertz, 2004].
The clinical features of PNDs are diverse and depend on the site of neurological involvement. PNDs can affect any part of the nervous system [Vernino, 2006]. Although a large variety of tumors have been reported in association with PNDs; adenocarcinoma of breast and ovary, SCLC, thymoma and Hodgkin disease are the most common tumors associated with PNDs. These syndromes occur in a very small subset of these patients.
Since PNDs are quite rare, very few formal therapeutic studies have been performed. In many cases, neurological symptoms respond poorly to current treatments. When approaching a patient with PND, the clinician’s main goals are to: (1) identify and eradicate the underlying malignancy as soon as possible, (2) identify those PNDs that are most likely to respond to immunotherapy, (3) counsel the patient and family about the nature of PNDs including the uncertainties and goals of treatment, and (4) initiate a treatment plan. The main advances in PNDs over the past decade have been in improved diagnostics and characterization of previously unrecognized syndromes. With respect to therapy for PNDs, there have only been a handful of controlled treatment trials to supplement a large literature of individual case reports. In this report, we briefly review the recognized paraneoplastic neurological syndromes and focus on the published experience with various treatment strategies.
Paraneoplastic neurological disorders are broadly defined as neurological signs and symptoms associated with malignancy, which are not explained by direct tumor invasion, metastasis, or iatrogenic causes such as chemotherapy or radiation. These disorders typically appear well before cancer is discovered, hence the initial diagnosis of PND is made by the neurologist rather than the oncologist. Several distinct clinical syndromes have been identified (Table 1), but many patients have a multifocal constellation of neurological signs and symptoms that does not perfectly match one of these syndromes. The clinical manifestations are never diagnostic, and a high index of suspicion is necessary for timely diagnosis. Similar neurological syndromes can occur from other causes including autoimmune disorders unrelated to cancer, cerebrovascular disease, atypical nervous system infections, toxic/metabolic conditions, or even inherited disorders. PNDs are rightly considered in the differential diagnosis of any otherwise unexplained neurological syndromes with a subacute onset and progressive course.
To assist in diagnosis, a number of serum antibodies reactive against neural tissue antigens have been identified (Table 2). These antibodies are strongly associated with cancer and have been detected unequivocally by several laboratories in a considerable number of patients with a wide range of neurological presentations [Graus et al. 2004]. Since different antibodies can be associated with similar clinical features, and any individual antibody test has low diagnostic sensitivity, paraneoplastic antibodies should generally be sought in appropriately designed panels that including several antibodies [De Beukelaar and Sillevis Smitt, 2006]. Also, it is important to remember that not all PND cases are associated with a detectable antibody marker [Darnell et al. 2003; Lawn et al. 2003]. Thus, a negative antibody test does not rule out PND, and further testing and management must be guided by clinical suspicion.
To provide some uniformity in clinical diagnosis, an international group in 2004 proposed diagnostic criteria for PNDs [Graus et al. 2004]. According to these criteria patients are divided to ‘definite’ and ‘possible’ groups (Box 1). Although these criteria are an improvement, many patients fail to reach the definition of ‘definite’ PND, and diagnostic uncertainty remains.
Computerized tomography scan of chest, abdomen and pelvis will uncover an underlying tumor in a large number of patients. If routine imaging is negative, in the face of high clinical suspicion or the presence of a predictive antibody, whole body fluorodeoxyglucose positron emission tomography (FDG-PET) should be considered. In prospective studies of patients with PNDs and well-characterized autoantibodies in whom conventional imaging was negative for underlying malignancy, whole body FDG-PET showed an abnormality in 37–83% [Linke et al. 2004; Younes-Mhenni et al. 2004; Rees et al. 2001]. However, a tissue diagnosis of cancer was not possible in many cases with positive FDG-PET imaging suggesting that false positives may occur. The authors conclude that FDG-PET should be reserved for patients with well-defined PNDs (usually those that are seropositive for a predictive paraneoplastic antibody) in whom conventional imaging fails to identify the tumor or where imaging results are equivocal. In some cases, especially those with anti-Yo or anti-Ma2 antibodies, PET scan can be negative [Castle et al. 2006], while surgical exploration may reveal small foci of cancer. Even when extensive radiologic diagnostic work up is negative, surgical exploration of the pelvis (in the case of anti-Yo antibody) or of the testes (in the case of anti-Ma2 antibody) may be justified [Mathew et al. 2007; Hetzel et al. 1990]. Conversely, when the clinical presentation is suggestive of PNDs (even in the absence of autoantibodies and negative initial imaging for the underlying tumor) it is a common practice to perform serial imaging studies (including FDG-PET when clinically indicated) at regular intervals for several years to monitor for the appearance of neoplasm [Basu and Alavi, 2008].
Several lines of evidence support the designation of PNDs as autoimmune disorders of the nervous system. The targets for most of the paraneoplastic antibodies are so-called ‘onconeuronal antigens’, proteins shared by both tumor cells and neural tissue constituents. Furthermore, pathological studies in PND cases (largely limbic encephalitis and cerebellar degeneration) have shown infiltration of the tumor as well as target nervous tissue by inflammatory cells [Rosenblum, 1993]. The inflammatory cells consist of perivascular accumulations of CD4+ T cells and B cells, as well as parenchymal CD8+ T cells and microglia [Bernal et al. 2002; Jean et al. 1994]. Both T cell and antibody-mediated processes have been implicated. Albert and co-workers showed that activated T cells from the CSF of a patient with paraneoplastic cerebellar degeneration could lyse target cells presenting Yo antigen in vitro [Albert et al. 1998]. Analysis of CSF frequently reveals mild lymphocytic pleocytosis, intrathecal synthesis of IgG and oligoclonal bands [Furneaux et al. 1990]. Paraneoplastic antibodies may be detected in the CSF, and in some cases, there is evidence of intrathecal synthesis of these antibodies.
The relative contributions of cell-mediated and humoral mechanisms to neural damage in PNDs are not clear and probably differ among different syndromes. In general, antibodies are more likely to play an important pathophysiological role in syndromes associated with antibodies against cell surface antigens [Darnell and Posner, 2003]. Examples include anti-voltage-gated calcium channel (VGCC) antibodies in Lambert–Eaton syndrome, anti-acetylcholine receptor antibodies in myasthenia gravis, and anti-voltage-gated potassium channel (VGKC) or anti-NMDA (N-methyl-D-aspartic acid) receptor antibodies in limbic encephalitis. In cases where the paraneoplastic antigens are located in the nucleus or cytoplasm, a pathogenic role for the corresponding antibodies has not been established [Sillevis Smitt et al. 1995]. In these cases, indirect (but persuasive) evidence points to cytotoxic cell-mediated immunity as the main mechanism [Tanaka et al. 2002; Benyahia et al. 1999; Tanaka et al. 1999].
The neuromuscular junction disorders (myasthenia gravis and Lambert–Eaton syndrome) represent special types of PND. Treatment strategies for those disorders have been described extensively and are not included as part of this discussion [Richman and Agius, 2003; McEvoy, 1994]. For other PNDs, there are no proven or accepted treatment protocols [Vedeler et al. 2006]. For the most part, treatment results have been disappointing [Posner and Dalmau, 1997; Grisold et al. 1995]. The mainstay of treatment is the early identification and elimination of the underlying tumor [Keime-Guibert et al. 1999]. These syndromes precede the identification of tumors in over 50% of cases [Darnell et al. 2003]. When found, tumors tend to be small and show only limited local metastasis. Therefore, cancer may be relatively more amenable to treatment [Rauer and Andreou, 2002; Maddison et al. 1999]. The search for malignancy needs to be exhaustive and complete. Early identification and management of the underlying tumor could lead to stabilization and in some cases improvement of the neurological syndrome. This appears to be the case with PNDs associated with anti-Hu, anti-CRMP5/CV2, anti-Ma2, but the neurological outcome for patients with cerebellar degeneration and anti-Yo antibodies is often poor despite optimal cancer treatment. If no cancer is found initially, the clinical suspicion should remain high and early repeat cancer screening should be scheduled. From a neurological standpoint, there are several potential beneficial effects of cancer treatment. First, it is important to eliminate the neoplasm as a stimulus for ongoing autoimmunity. Second, the immunosuppressive effects of chemotherapeutic agents may help dampen the immune response. In the case of POEMS syndrome, treatment of the plasmacytoma (with radiation) can result in improvement of neuropathy, perhaps by eliminating the malignant cells that are producing pathogenic mediators, most notably vascular endothelial growth factor [Dispenzieri and Gertz, 2004].
Immunomodulatory therapy can be considered in cases where a malignancy has not been identified, in cases where oncological treatment has been completed, or in conjunction with cancer treatment. Steroids, intravenous immunoglobulin (IVIG), plasma exchange (PLEX), rituximab, and cyclophosphamide are therapies that have been reported Table 3. Data on treatment response is largely based on small retrospective studies and case reports. In an earlier review of available retrospective case series, only 33 cases of effective treatment were documented out of 259 reported cases [Posner and Dalmau, 1997; Grisold et al. 1995].
There have been no placebo-controlled clinical trials. A few systematic prospective series have been reported. Two studies were conducted using IVIG alone or IVIG in combination with pulse intravenous cyclophosphamide and methylprednisolone [Keime-Guibert et al. 2000; Uchuya et al. 1996]. Treatment, however, was given for a variable duration and in combination with chemotherapy in many cases. Among patients with progressive neurological disease, 35–40% of patients stabilized neurologically, and only one patient improved. The authors concluded that this immunomodulatory treatment was not useful for patients with severe disability but may provide a useful stabilization of disability in patients who are still ambulatory [Keime-Guibert et al. 2000].
In a single-center, prospective, open-label study to evaluate the efficacy of initial PLEX in combination with either chemotherapy or oral cyclophosphamide [Vernino et al. 2004], a positive response was seen in 45% of the subjects (measured as improvement in Rankin disability score and in activities of daily living) after 6 months. Most of the stabilization and recovery occurred gradually, so the beneficial effects of PLEX were uncertain. Hematological side effects of cyclophosphamide were a problem in many cases. The conclusion was that immunosuppressive therapy should be considered early in the course of PNDs, even when there is no evidence of active malignancy. Similar to the aforementioned IVIG studies, patient with less severe disability and those with peripheral, as opposed to central, neurological syndromes tended to have a better outcome [Rosenfeld and Dalmau, 2006; Vernino et al. 2004; Keime-Guibert et al. 2000].
It appears that paraneoplastic disorders in which antibodies are directed against cell surface antigens are more amenable to immunomodulatory treatment. In these syndromes, antibodies may produce a functional neuronal deficit rather than a cytotoxic immune response which would irreversibly damage neurons. This includes the neuromuscular syndromes (Lambert–Eaton syndrome, myasthenia gravis and neuromyotonia), paraneoplastic cerebellar degeneration with antibodies directed against metabotropic glutamate receptors (mGluR1), and paraneoplastic encephalitis associated with anti-NMDA receptor antibodies [Dalmau et al. 2007; Shams’ili et al. 2003]. The latter disorder is an important (and relatively recent) discovery. The patients, usually young women with occult ovarian teratoma, present with dramatic behavioral changes, dyskinesias and sometimes respiratory failure. The disease may be fatal if undiagnosed, but the prognosis is good if the teratoma is removed and immunomodulatory therapy is initiated. Detection of antibodies against NMDA receptors in the serum or CSF confirm the diagnosis [Dalmau et al. 2007]. Of note, there is a subset of patients with autoimmune limbic encephalitis who harbor antibodies directed against neuronal cell membrane antigens including NMDA, VGKC, and yet uncharacterized neuropil antigens in the hippocampus and cerebellum. The neurological syndromes in these patients are less frequently associated with cancer and respond better to immunotherapy as opposed to those associated with neuronal nuclear or cytoplasmic antibodies (such as anti-Hu) [Tüzün and Dalmau, 2007].
In other PNDs, the antibodies are directed against intracellular neuronal antigens. In these cases, the antibodies appear to be a marker of a cell-mediated immune response against the cancer cells. These disorders are often characterized by selective neuronal loss and less potential for neurological recovery. The goals of treatment are to halt the pathological autoimmune process. Treatment of the underlying malignancy with or without additional therapy to suppress cellular immunity seems appropriate. However, it should be remembered that concurrent use of chemotherapeutic agents and immunosuppressive agents can lead to increased toxicity [Rosenfeld and Dalmau, 2006]. Since the immune response likely contributes to suppressing cancer growth and metastases, immunosuppression carries a theoretical risk of more malignant tumor behavior.
Selection of an immunomodulatory agent for the treatment of PND should be based on the individual syndrome and clinical situation. There are a variety of options, and dosing generally follows clinical practice for other autoimmune neurological conditions (such as multiple sclerosis, vasculitis, or inflammatory demyelinating neuropathy). Although there are case reports of benefit for all these agents, the efficacy is still unknown. Potential risks (including the cost of treatment) must be weighed against an uncertain benefit.
Steroids (intravenous methylprednisolone and oral prednisone) alone or in combination with other immunosuppressive agents, have been used extensively in the treatment of PNDs [Gultekin et al. 2000; Keime-Guibert et al. 2000; Oh et al. 1997]. There is no consensus on standard dose of steroids for treatment of PNDs. In one study, intravenous methylprednisolone was used in combination with other immunomodulatory treatments at 1g per day for 3 days [Keime-Guibert et al. 2000]. Others have reported therapeutic responses with intravenous methylprednisolone at 500mg/day for 5 days or oral prednisone at 1mg/kg/day. Steroids may have a particular role in rapidly progressive cases with evidence of CNS inflammation (e.g. edema, contrast enhancement, CSF pleocytosis).
Intravenous immunoglobulin has been used and reported in a variety of paraneoplastic disorders [Phuphanich and Brock, 2007; Widdess-Walsh et al. 2003; Gultekin et al. 2000; Keime-Guibert et al. 2000; Mowzoon and Bradley, 2000; Blaes et al. 1999; Guy and Aptsiauri, 1999; Oh et al. 1997; David et al. 1996; Counsell et al. 1994; Glantz et al. 1994; Moll et al. 1993]. Dosing varies from maintenance doses of 0.4g/kg/day to initial doses of 2g/kg. The mechanism of action of IVIG in PNDs is uncertain. Based on experience with other neurological disorders, IVIG seems most appropriate for PNDs affecting the peripheral nervous system and those that may be antibody-mediated.
Similarly, plasma exchange reduces serum levels of antibodies, cytokines, and other inflammatory mediators and would be expected to be most useful for PNDs directly mediated by antibodies [Rosenfeld and Dalmau, 2006]. There are numerous individual case reports of clinical improvement using PLEX [Taniguchi et al. 2006; Armstrong et al. 2005; Rickman et al. 2000; David et al. 1996; Weissman and Gottschall, 1989; Cocconi et al. 1985], as well as one prospective study [Vernino et al. 2004]. Furthermore, partial or transient improvement was observed in 75% of PND patients treated with immunoadsorption therapy in a prospective therapy [Batchelor et al. 1998]. The number and timing of PLEX treatments varies among reports, but a common protocol is five or six exchanges performed on alternate days [Rosenfeld and Dalmau, 2006; Vernino et al. 2004]. Contraindications include coagulopathy and heart failure. Potential complications of PLEX are cardiac arrhythmias, hypotension and infection at the site of catheter, sepsis, pneumothorax and electrolyte abnormalities.
Cyclophosphamide has been commonly used to treat PNDs [Vernino et al. 2004; Mowzoon and Bradley, 2000; Faris et al. 1998, Schmierer et al. 1998; Stark et al. 1995; Batson et al. 1992; Oh et al. 1991; Bruyland et al. 1984]. It is an alkylating agent that is converted to active metabolites by the liver. Cyclophosphamide is often used when other immunomodulatory treatments fail to control neurological deterioration. Responses have been reported with intermittent intravenous dosing ranging from 500mg to 1g/m2 administered monthly [Rosenfeld and Dalmau, 2006]. Others have used daily oral cyclophosphamide at 2mg/kg/day [Vernino et al. 2004]. Side effects include nausea, vomiting, headache, myelosuppresion, hemorrhagic cystitis and increased risk of lympho- and myeloproliferative disorders. In some cases, cyclophosphamide can be incorporated into a cancer treatment protocol, but in most cases, this agent would not be appropriate in a patient receiving chemotherapy because of cumulative side effects.
Other oral immunosuppressants could be considered and may be more compatible with ongoing cancer treatment. The data on use of tacrolimus (FK506) in PND are quite limited. This potent inhibitor of lymphocyte proliferation is more commonly used to prevent rejection in organ transplant patients. Albert and co-workers used tacrolimus at a dose of 0.15mg/kg/day for 14 days, followed by 0.3mg/kg/day for 7 days to treat paraneoplastic cerebellar degeneration [Albert et al. 2000]. Although, this regimen decreased the number of activated T cells in the cerebrospinal fluid, no clinical improvement was observed. Mycophenolate mofetil, a selective inhibitor of lymphocyte proliferation, is associated with less myelosuppression than cyclophosphamide and anecdotally appears to be safe in PND patients taking chemotherapy. However, no data on efficacy are currently available.
Rituximab is a monoclonal antibody directed against CD20 molecule on the surface of B cells, which has proved effective in treatment of non-Hodgkin lymphomas. Recent uncontrolled studies and case reports suggest usefulness of this drug in treating PNDs [Bell et al. 2008; Esposito et al. 2008; Shams’ili et al. 2006]. In one study, a maximum of four monthly infusions of rituximab were used at 375mg/m2 in nine patients with PND [Shams’ili et al. 2006]. Three patients showed clinical improvement. More data are needed to determine whether a specific subset of PND patients might benefit from rituximab. A number of other monoclonal antibody therapies have been developed recently as immunomodulatory treatments for multiple sclerosis and other systemic autoimmune diseases. Eventually, some of these biologicals may prove useful in PNDs.
Current therapeutic modalities for paraneoplastic disorders are, for the most part, based on anecdotal case reports and uncontrolled small case series. Prospective studies of treatment efficacy in PNDs have been hampered by the heterogeneity of the clinical presentation as well as the rarity of these syndromes. The best approach in treatment of any paraneoplastic disorder is to identify and treat underlying tumor and at the same time provide appropriate supportive care and symptomatic therapies. Recent advances have improved our ability to recognize PNDs. These include additional clinical descriptions of new paraneoplastic syndromes, an increase in the number and variety of diagnostic antibody markers, and improvements in cancer imaging (FDG-PET).
Immunomodulatory treatments for PND are gaining acceptance, even though few prospective studies have been performed. With improved awareness, earlier diagnosis and treatment seems increasingly possible. Improved therapeutics will result not only from new agents but also systematic early institution of treatment. Since PNDs are rare, patient registries and multicenter prospective treatment studies will be important components for future progress.
Dr Vernino has served as a consultant for Athena Diagnostics.