A prospective open-label study of sirolimus for the treatment of anti-Hu associated paraneoplastic neurological syndromes

Abstract

Background

Several lines of evidence suggest a T cell–mediated immune response in paraneoplastic neurological syndromes with anti-Hu antibodies (Hu-PNS). In order to investigate whether suppression of T cell–mediated immune responses in Hu-PNS patients improved their neurological outcome, we performed a prospective open-label, single-arm study on sirolimus.

Methods

Seventeen progressive Hu-PNS patients were treated with sirolimus with an intended treatment duration of 8 weeks. Primary outcome measures were (i) functional improvement, defined as a decrease of one or more points on the modified Rankin Scale (mRS), and (ii) improvement of neurological impairment, defined as an increase of one or more points on the Edinburgh Functional Impairment Tests (EFIT).

Results

One patient showed improvement on both clinical scales (mRS and EFIT). This patient presented with limbic encephalitis and improved dramatically from an mRS score of 3 to mRS 1. Another patient, with subacute sensory neuronopathy, remained stable at mRS 2 and improved one point on the EFIT scale. The other patients showed no improvement on the primary outcome measures. Median survival was 21 months.

Conclusion

We conclude that treatment of Hu-PNS patients with sirolimus may improve or stabilize their functional disabilities and neurological impairments. However, the effects of this T cell–targeted therapy were not better than reported in trials on other immunotherapies for Hu-PNS.

Trial Registration https://www.clinicaltrialsregister.eu/ctr-search/trial/2008-000793-20/NL.

Among paraneoplastic neurological syndromes (PNS) with well-characterized onconeural antibodies, anti-Hu antibody–associated PNS (Hu-PNS) are the most frequent.^1,2 Hu-PNS have a very poor prognosis: more than half of the patients become bedridden (modified Rankin Scale [mRS] score ≥4), only 2%–6% of patients improve, and median survival is <12 months.^3,4 Hu-PNS are thought to result from an immune response against Hu-expressing tumor cells that additionally destroys Hu-expressing neurons.⁵ There are several reasons why neuronal destruction in Hu-PNS is more likely caused by T cells than anti-Hu antibodies (Hu-Abs): Hu proteins are intracellular proteins,⁶ animal models failed to demonstrate Hu-Ab–induced neurological disease,⁷ and autopsy studies showed T cell infiltrates with frequent appositions of cytotoxic T cells to neurons with associated neuronal loss.^8–10

Sirolimus (Rapamune) is an immunosuppressive drug that specifically inhibits activated T cells.¹¹ Sirolimus binds to FKBP-12, an intracellular protein, to form an immunosuppressive complex that inhibits the regulatory kinase mammalian target of rapamycin.¹¹ This inhibition suppresses cytokine-mediated T cell proliferation. As a result, T cells cannot proliferate following antigenic stimulation, and T cell–mediated immune responses are strongly suppressed. In addition, sirolimus has anti-oncogenic and anti-angiogenic effects that may prevent growth of the underlying tumor.¹²

We hypothesized that cellular immune suppression with sirolimus could stop neuronal damage by the presumed T cell–mediated immune response in Hu-PNS and may improve neurological outcome. To test this hypothesis, we prospectively treated 17 progressive Hu-PNS patients with sirolimus and recorded neurological function and impairment using well-defined clinical scales in a prospective open-label, single-arm study.

Methods

Eligibility Criteria

Inclusion criteria included high serum titers of Hu-Ab (≥400 by indirect immunofluorescence, confirmed by western blotting) and progression of neurological symptoms, defined as neurological deterioration over the last 4 weeks. From May 2008 to October 2012, we identified 68 patients with high-titer Hu-Ab. Four of these patients died before test results became available. Eighteen patients did not meet the inclusion criteria according to their treating physician: 12 had no progressive disease over the last 4 weeks, while 6 patients had other potential causes for their neurological symptoms (toxic neuropathy or brain metastasis). Of the remaining 46 patients who were invited to participate in this study, 29 declined participation (mainly because of severe illness or unwillingness to travel to our institution) or did not respond to our invitation. Seventeen patients gave written informed consent and participated in this trial. Clinical characteristics of the 17 included patients are summarized in Table 1. All patients were evaluated throughout the study by 1 of 2 clinical investigators (J.E.B., P.A.S.S.). The study was approved by the Erasmus University Medical Center Institutional Review Board.

Table 1.

Patient characteristics

No.	Age, y	Sex	PNS	Symptoms to Diagnosis (mo)	Tumor	Symptoms to Tumor (mo)	Tumor Stage	Tumor Treatment	Tumor Response	Symptoms to Last FU (mo)	Dead/Alive	Cause of Death
1	68	M	PEM (LE)	7	SCLC	8	ED	Chemo + RT	PD	27	Dead	Tumor
2	67	M	BE	9	SCC parotid	−8	T1N2bM0	Chemo + RT + surg	PD	16	Dead	PNS
3	77	F	PEM (SSN), LEMS	4	SCLC	6	LD	No	NA	8	Dead	PNS
4	60	M	PCD	4	Noa	NA	NA	No	NA	10	Dead	PNS
5	52	F	SSN	11	SCLC	−4	LD	Chemo + RT	CR	43	Alive	NA
6	75	F	PCD	2	SCLC	4	LD	Chemo + RT	CR	33	Alive	NA
7	46	F	PEM (SSN)	1	Lungb	1	ED	No	NA	2	Dead	PNS
8	66	F	PLE	7	Lungc	8	LD	No	NA	21	Dead	PNS
9	77	F	SSN	5	Nod	NA	NA	No	NA	24	Alive	NA
10	73	F	SSN	2	SCLC	4	LD	Chemo + RT	CR	21	Alive	NA
11	65	F	SSN, AN	4	Chondrosarcoma	−312	Metastatic	No	NA	14	Dead	PNS
12	59	F	MN	2	SCLC	−24	LD	Chemo + RT	CR	9	Dead	PNS
13	64	F	SSN	0	SCLC	−4	LD	Chemo + RT	CR	16	Alive	NA
14	64	M	PLE	0	SCLC	1	LD	Chemo + RT	PR	7	Dead	Tumor
15	62	M	SSN	2	SCLC	2	LD	Chemo + RT	CR	11	Alive	NA
16	66	M	PCD	4	Lunge	5	LD	No	NA	10	Alive	NA
17	67	F	SSN	1	SCLC	2	LD	Chemo	CR	7	Alive	NA

Abbreviations: FU, follow-up; PEM, paraneoplastic encephalomyelitis; BE, brainstem encephalitis; LEMS, Lambert–Eaton myasthenic syndrome; PCD, paraneoplastic cerebellar degeneration; SSN, subacute sensory neuronopathy; AN, autonomic neuropathy; PLE, paraneoplastic limbic encephalitits; MN, motor neuronopathy; SCLC, small-cell lung cancer; SCC parotid, small-cell cancer of parotid gland; NA, not applicable; ED, extensive disease; LD, limited disease; Chemo, chemotherapy; RT, radiotherapy; surg, surgery; PD, progressive disease; CR, complete response; PR, partial response. In PEM patients, the predominant syndromes are indicated in parentheses.

^aCT-thorax and 2-fluoro-2-deoxy-d-glucose–PET-CT scan negative.

^bExtensive mediastinal and pleural lesions on CT scan.

^cMediastinal 2-fluoro-2-deoxy-d-glucose positive lymph nodes on PET-CT scan.

^dCT-thorax/abdomen unchanged.

^eMediastinal 2-fluoro-2-deoxy-d-glucose positive lymph nodes on PET-CT scan (not accessible for biopsy).

Treatment Plan

Sirolimus treatment was initiated with an oral loading dose of 6 mg sirolimus per day for 3 consecutive days, followed by oral maintenance dosing of 3 mg/d. The dosing was adjusted weekly to maintain trough concentrations of 8–12 ng/mL. If the patient was under cotherapy with drugs that induce or inhibit the activity of CYP3A4, the sirolimus dose was adjusted accordingly, under the guidance of the clinical pharmacologist. Patients were informed that any planned change in comedication needed to be reported to the treating physician, in order to avoid adverse events due to drug-drug interactions.

Laboratory Tests

Serum and CSF were sampled at baseline and after 4 weeks of sirolimus treatment. Serum was additionally sampled at weeks 8 and 16. Sirolimus blood trough levels were determined in the clinical chemistry laboratory (target level 8–12 ng/mL) using a chromatographic assay. IgG titers of the Hu-Ab were determined as described.³

Assessment of Efficacy

The primary endpoints of the study were functional and neurological improvement after 8 weeks of sirolimus. Functional improvement was defined as a decrease of one point or more on the mRS compared with the baseline evaluation. Improvement of neurological impairment was assessed with the Edinburgh Functional Impairment Tests (EFIT), which incorporate objective measures of upper and lower limb function, memory, and a rating scale for dysphasia.¹³ Overall EFIT = 0 indicates no change, EFIT > 0 indicates significant neurological improvement, and EFIT < 0 indicates significant neurological deterioration.

Secondary endpoints included reduction of CSF protein and white blood cells (WBCs), decrease in Hu-Ab titers in serum and CSF, and improvement in activities of daily living as evaluated by the Barthel index.¹⁴ We added a post-hoc analysis to enable comparison with previous studies. In this post-hoc analysis, outcome was considered “successful” when a patient with an mRS score ≤3 improved or stabilized (ie, remained ambulatory) and when a patient with an mRS score ≥4 (bedridden patient) improved to ≤3 (ambulatory) after the 8th week of sirolimus treatment compared with baseline, as defined by Keime-Guibert et al.¹⁵

Statistical Methods

We compared WBC, total protein concentration, and Hu-Ab titers in baseline CSF and CSF obtained after 4 weeks of treatment by means of the Wilcoxon matched pairs test. We used the same test to compare Hu-Ab serum titers at baseline and end of study. P-values were 2-sided and a significance level of α = .05 was used. All statistical analyses were performed using GraphPad Prism version 5 software.

Results

Patient Flow and Treatment

A total of 17 patients started treatment with sirolimus, 11 of whom completed the full 8 weeks of treatment (Table 2). Six of these 11 patients received sirolimus for 9–11 weeks because of low sirolimus blood levels during the titration phase. In 6 patients, sirolimus treatment was terminated before 8 weeks. In 2 of these patients, serious adverse events related to sirolimus were reported (patients 8 and 15; see below). In 2 patients, withdrawal was requested for fear of increased hematological toxicity by oncologists in other hospitals who wanted to start treatment with chemotherapy. In 1 patient, sirolimus was withdrawn after 4 weeks because of obvious neurological deterioration, and 1 patient died after 1 week of treatment of the neurological disorder. The range of the average daily sirolimus dose was 2.3–5.3 mg/d, resulting in average trough levels of 3.0–11.7 µg/L (Table 2).

Table 2.

Sirolimus treatment duration and intensity

No.	Sirolimus			Reason Early Termination	Concurrent Chemotherapy
	Wk	Average Daily Dose, mg	Average Trough Level, μg/L
1	8	3.0	7.1	NA	Yes
2	3	3.0	3.1	Initiation chemotherapy	No
3	9	3.0	4.8	NA	No
4	8	5.0	3.4	NA	No
5	9	3.0	5.2	NA	Yes
6	5	3.8	3.9	Neurological deterioration	No
7	1	3.0		Died of PNS	No
8	4	3.8	4.1	SAE, severe epistaxis	No
9	9	2.3	11.5	NA	No
10	3	4.7	5.5	Initiation chemotherapy	No
11	9	5.3	7.4	NA	No
12	11	4.8	6.1	NA	No
13	8	4.1	10.2	NA	No
14	10	4.7	8.0	NA	Yes
15	2	3.0	4.2	SAE, severe erythrodermia	Yes
16	8	2.3	8.9	NA	No
17	8	3.4	11.6	NA	Yes

Abbreviations: SAE, serious adverse event; NA, not applicable.

Toxicity

As stated above, 2 patients were unable to complete the treatment because of serious adverse events. Patient 8 developed severe epistaxis that caused a drop in hemoglobin levels (from 5.3 to 4.8 mmol/L) requiring hospitalization and transfusion of 2 units of packed red cells. Sirolimus was stopped and the patient recovered completely (hemoglobin 6.0 mmol/L). Patient 15 developed a generalized erythrodermia for which the consultation of a dermatologist was necessary. Sirolimus was withdrawn and topical corticosteroids were prescribed. Two weeks later, the patient was seen again by the dermatologist, and the skin was recovering. One week later, the patient was admitted to the hospital because of a Staphylococcus aureus bacteremia with the skin as possible porte d'entrée. The patient recovered completely from the erythrodermia and the bacteremia. Patient 1 developed oral mucositis, thrombocytopenia, and hypokalemia, from which he recovered completely.

Primary Endpoints

Only in patient 14 did the mRS score improve (from 3 to 1) with concomitant improvement of one point on the EFIT score (Table 3). This patient suffered from limbic encephalitis (memory deficits, sexual disinhibition, and visual hallucinations) and received concomitant treatment with chemotherapy for the underlying small-cell lung cancer. After treatment, his memory and behavior improved dramatically, and he had no more hallucinations. Patient 5, with subacute sensory neuronopathy, improved one point on the EFIT scale while remaining stable at an mRS score of 2. Overall, 10 patients had stable mRS scores (59%), while 6 patients (35%) showed further functional deterioration. The neurological outcome (EFIT) was stable in 7 of 14 evaluable patients (50%) and deteriorated in 5 (36%).

Table 3.

Primary outcome measures

No.	Sirolimus, wk	Concurrent Chemotherapy	Functional Outcome	mRS Baseline	mRS Change	Neurological Outcome	EFIT Baseline	EFIT Overall
1	8	Yes	Stable	3	0	Stable	2	0
2	3	No	Worse	2	1	Worse	0	−1
3	9	No	Worse	3	2	Worse	3	−1
4	8	No	Stable	3	0	Worse	4	−1
5	9	Yes	Stable	2	0	Improved	2	1
6	5	No	Worse	3	1	Worse	2	−1
7	1	No	Worse	3	3	NEa	2
8	4	No	Stable	3	0	Stableb	2	0
9	9	No	Stable	4	0	Stable	2	0
10	3	No	Stable	3	0	NEc	1
11	9	No	Stable	4	0	Stable	2	0
12	11	No	Worse	3	1	Stable	2	0
13	8	No	Stable	4	0	Stable	3	0
14	10	Yes	Improved	3	−2	Improved	2	1
15	2	Yes	Stable	3	0	NEc	1
16	8	No	Worse	4	1	Worse	4	−2
17	8	Yes	Stable	5	0	Stable	2	0

Abbreviation: NE, not evaluable.

^aPatient died because of progressive PNS.

^bPatient's PNS remained stable on sirolimus. After discontinuation because of side effects, her PNS progressed.

^cEFIT score was not determined, mRS score was determined by contacting patient and treating physician.

Functional and neurological outcome after ≤8 weeks of treatment with sirolimus. An mRS change <0 indicated functional improvement; an mRS change of 0 indicated stable functional outcome; an mRS change >0 indicated functional deterioration. An overall EFIT of >0 indicated neurological improvement; an overall EFIT of 0 indicated stable neurological function; an overall EFIT <0 indicated neurological progression.

Secondary Endpoints

Improvement in activities of daily living (a rise of 5 points or more on the Barthel index) was seen in patients 11 (+15) and 13 (+20). Both patients had a stable mRS score of 4 and unchanged EFIT score. Laboratory evaluations showed in CSF a median of 5 WBC/µL at baseline versus 5 WBC/µL after 4 weeks of treatment (P = .38, n = 12). The median CSF protein concentration was 0.42 g/L at baseline and 0.35 g/L after 4 weeks (P = .12). The median CSF Hu-Ab titer changed from 32 at baseline to 64 at week 4 (P = .47). In serum, the median Hu-Ab titer at baseline was 3200, at 4 weeks 2400 (P = .16 vs baseline, n = 12), at 8 weeks 1600 (P = .03 vs baseline, n = 11), and at 16 weeks 800 (P = .50 vs baseline, n = 5).

Post-hoc Analyses

According to the criteria of Keime-Guibert and colleagues, treatment response would be classified as successful in 7 patients (41%; patients 1, 4, 5, 8, 10, 14, and 15).¹⁵ Median survival was 21 months.

Discussion

One patient showed a dramatic clinical improvement during sirolimus treatment, with a decrease of 2 points on the mRS (from 3 to 1) and concomitant improvement on the overall EFIT of +1. Of the remaining patients, 59% had a stable functional outcome, while 35% showed further deterioration. Treatment response was successful in 7 patients (41%).¹⁵

Six other studies have reported treatment results in Hu-PNS patients using the mRS as the primary outcome measure, and have recently been reviewed elsewhere.¹⁶ In 3 retrospective studies, treatment with plasma exchange and methylprednisolone with or without cyclophosphamide was successful in 0 of 6 (0%),¹⁷ intravenous immunoglobulin treatment was successful in 6 of 17 (35%),¹⁸ while treatment with intravenous immunoglobulin, cyclophosphamide, and methylprednisolone was successful in 2 of 9 (22%)¹⁵ evaluable Hu-PNS patients. In a prospective study, treatment with plasma exchange combined with either cyclophosphamide or chemotherapy (in patients with a tumor) was successful in 3 of 5 (60%) Hu-PNS patients.¹⁹ In 2 prospective studies, rituximab was successful in 4 of 8 (50%)²⁰ and human chorionic gonadotropin in 7 of 15 (47%)²¹ Hu-PNS patients. The rate of success in the current study is within the range of the success rates of previous studies.

The current study suggests that immunosuppressive and immuomodulatory therapy may modify the course of Hu-PNS. However, several confounding factors may have contributed to these results. The patient who showed improvement on both primary outcome measures (patient 14) suffered limbic encephalitis, a syndrome that appears most responsive to treatment and can rarely spontaneously resolve, even in the presence of Hu-Abs.³ In addition, this patient received concomitant chemotherapy, and several studies have demonstrated that effective treatment of the tumor is important to at least stabilize Hu-PNS.^3,4 Four of the other 6 patients with a “successful” outcome received chemotherapy, with a complete response in 3 of them. Patient 5 had a long interval from symptom onset to diagnosis (11 mo), which is also associated with a more favorable outcome.³

During sirolimus treatment, we did not find significant changes in the CSF WBC count, protein concentration, or Hu-Ab titers. Also, serum titers were not significantly affected by sirolimus. The significant decrease in the serum titer at 8 weeks (P = .03) probably reflects multiple testing.

To summarize, this study and other studies suggest that immunomodulation may modify the course of Hu-PNS. However, we can currently not conclude that T cell–targeted therapies should be preferred above other immunotherapies, and more effective treatments are clearly needed.

Funding

This work was supported by the Gratama Foundation (Harlingen, the Netherlands).