Table e-1: Criteria for Inclusion
  • Ambulatory subjects with extremity weakness and/or spasticity due to ALS.

  • Have the ability to understand the requirements of the study, provide written informed consent, understand and provide written authorization for the use and disclosure of Protected Health Information (PHI) [per Health Insurance Portability and Accountability Act (HIPAA) Privacy Ruling] and comply with the study procedures.

  • Subjects with sporadic or familial ALS, meeting the definition of laboratory-supported probable, probable or definite ALS according to the World Federation of Neurology El Escorial Criteria. At the time of consent subjects should be within 24 months of symptom onset. At the time of surgery, subjects should be within 36 months of symptom onset.

  • Age 18 years or older.

  • Females must have a negative serum pregnancy test and practice an acceptable method of contraception or be of non-childbearing potential (post-menopausal for at least 2 years or surgically sterile [hysterectomy, oophorectomy or surgical sterilization]).

  • Geographic accessibility to the study center and the ability to travel to the clinic for study visits.

  • Presence of a willing and able caregiver.

  • Medically able to undergo cervical and lumbar (Group E only) laminectomy or laminoplasty as determined by the site Principal Investigator and neurosurgeon.

  • Subjects with positive toxoplasmosis IgG agrees to take Bactrim DS 1 tab PO 3 times weekly during the 6-month treatment period

  • Medically able to tolerate the immunosuppression regimen consisting of basiliximab, tacrolimus, mycophenolate mofetil, prednisone and methylprednisolone as determined by the site PI.

  • Agrees to the visit schedule as outlined in the informed consent.

  • Not taking riluzole (Rilutek®) or on a stable dose for ≥ 30 days prior to surgery.

  • Vital capacity ≥ 60% of predicted normal for age, height and gender measured in the seated position and ≥50% in supine position at screening and at ≤ 7 days prior to surgery.

Details of statistical analysis:

Let Yijk denote the response of patient j in groupi at time tk, where j = 1 if in the control group and 0 otherwise. Then our model is as follows:

where β0 is the overall Y intercept, β0i is the fixed effect on the mean for group I, which can be interpreted as the increment to the Y intercept for group I;Zj(i) is an indicator variable equaling 1 if patient j is in the control group and 0 otherwise, γj(i) is the random effect for patient j in group I;γj(i) is distributed as a normal random variable with mean of 0 and standard deviation of σg, β1 is the overall slope fixed effect, β1i is the fixed effect on the slope for group I and can be interpreted as the increment to the slope for group I, and εijk is the random error associated with patient j in group I at time tk, ε is distributed as a normal random variable with mean of 0 and standard deviation of σe. We assume γj(i) and εijk are independent and that the covariance matrix has a variance components structure, i.e., for the covariance element in row i, column j of the matrix, , where i corresponds to the kth effect, and I{i=j} is the indicator function equaling 1 when i=j and 0 otherwise. We use F tests to assess the following null hypotheses:

H0:β1i = 0 (determining if there is a difference between groups in the decline over time), H0:β1 = 0 (determining if there is a significant decline over time), and H0:β0i = 0 (determining if there is a difference in the Y intercept for the two groups). In the tables we represent these terms as follows: estimated β0 = intercept, estimated β0i = Control, estimated β1 = time, and estimated β1i = Control*Time. For each outcome, the full model (shown above) was analyzed, and a formal test of the third H0 above was conducted. If not significant, a reduced model (setting β1i = 0) was fit. In this case, the parameter estimates and associated standard errors and p-values are presented, while the non-significance of the Control*Time term is represented as “---“ for both the estimate (standard error) as well as for the associated p-value, indicating that the simpler model was fit. In addition, the p-value given by the statistical procedure for the Intercept term is derived from the test of the hypothesis that the true value of the intercept is equal to 0. Since this is not realistic for these data, the p-value for this term is also represented as “---“.

Table e-2. Parameter estimates from mixed models analysis of change in outcomes over time, cases v EU controls. Only observations post-intervention to day 270 are used.

Comparison  / Phase 1+2 vs EU-1+2 / Phase 2 vs EU-2
Outcome / Parameter / Estimate (SE) / p / Estimate (SE) / p
ALSFRSR
Intercept / 35.62 (1.46) / --- / 37.56 (1.67) / ---
Controla / -0.079 (1.52) / 0.9588 / -1.05 (1.80) / 0.5621
Timeb / -0.030 (0.010) / 0.0042 / -0.043 (0.012) / 0.0011
Control*Timec / --- / --- / --- / ---
FVC % Predicted
Intercept / 81.86 (3.81) / --- / 81.24 (4.22) / ---
Controla / -5.78 (4.10) / 0.1645 / -5.96 (4.51) / 0.1964
Timeb / -0.004 (0.027) / 0.8823 / -0.002 (0.030) / 0.9564
Control*Timec / --- / --- / --- / ---

aControl indicates the difference in outcome value at time of intervention for the control group relative to the study group; bTime indicates the rate of outcome value change per day since (first) intervention; cControl*Time indicates whether the control group subjects subjects change at a different rate than trial subjects. Use of “---“ indicates that a model with the term was evaluated, it was not statistically significant, and thus the estimates and p-values from the model without the term are presented.

Table e-3. Parameter estimates from mixed models analysis of change in outcomes over time, cases v CEF placebo group. Only observations post-intervention are used. In this case, only values up to 270 days post-intervention are used.

Comparison  / Phase 1+2 vs CEF / Phase 2 vs CEF
Outcome / Parameter / Estimate (SE) / p / Estimate (SE) / p
ALSFRSR
Intercept / 36.40 (0.98) / --- / 37.10 (1.19) / ---
Controla / -0.10 (0.92) / 0.9104 / -0.61 (1.16) / 0.6007
Timeb / -0.037 (0.004) / <0.0001 / -0.038 (0.004) / <0.0001
Control*Timec / --- / --- / --- / ---
FVC % Predicted
Intercept / 88.38 (2.82) / --- / 88.09 (3.57) / ---
Controla / -3.33 (2.68) / 0.2161 / -2.80 (3.46) / 0.4200
Timeb / -0.064 (0.013) / <0.0001 / -0.066 (0.013) / <0.0001
Control*Timec / --- / --- / --- / ---
Maximum Grip Strength
Intercept / 51.21 (3.71) / --- / 43.98 (4.44) / ---
Controla / -9.82 (3.47) / 0.0052 / -1.65 (4.27) / 0.6992
Timeb / -0.048 (0.017) / 0.0066 / -0.055 (0.018) / 0.0020
Control*Timec / --- / --- / --- / ---

aControl indicates the difference in outcome value at time of intervention for the control group relative to the study group; bTime indicates the rate of outcome value change per day since (first) intervention; cControl*Time indicates whether the control group subjects subjects change at a different rate than trial subjects. Use of “---“ indicates that a model with the term was evaluated, it was not statistically significant, and thus the estimates and p-values from the model without the term are presented.

Table e-4. Parameter estimates from mixed models analysis of change in outcomes over time, cases v ProACT. Only observations post-intervention are used. In this case, only values up to 270 days post-intervention are used.

Comparison  / Phase 1+2vsProACT / Phase 2vsProACT
Outcome / Parameter / Estimate (SE) / p / Estimate (SE) / p
ALSFRSR
Intercept / 35.36 (0.78) / --- / 35.99 (0.99) / ---
Controla / 3.34 (0.77) / <0.0001 / 2.73 (0.99) / 0.0059
Timeb / -0.028 (0.002) / < 0.0001 / -0.028 (0.002) / <0.0001
Control*Timec / --- / --- / --- / ---
FVC % Predicted
Intercept / 87.82 (3.20) / --- / 87.34 (4.31) / ---
Controla / -11.10 (3.19) / 0.0005 / -10.59 (430) / 0.0141
Timeb / -0.059 (0.006) / <0.0001 / -0.059 (0.006) / <0.0001
Control*Timec / --- / --- / --- / ---

aControl indicates the difference in outcome value at time of intervention for the control group relative to the study group; bTime indicates the rate of outcome value change per day since (first) intervention; cControl*Time indicates whether the control group subjects subjects change at a different rate than trial subjects. Use of “---“ indicates that a model with the term was evaluated, it was not statistically significant, and thus the estimates and p-values from the model without the term are presented.

Dataset / Variable / Lower CL / Upper CL
Cef / ALSFRSr / -0.0471 / -0.0358
FVC / -0.1093 / -0.0728
Max GST / -0.0798 / -0.0448
Pro-ACT / ALSFRSr / -0.0375 / -0.0302
FVC / -0.1251 / -0.0720
EU-1+2 / ALSFRSr / -0.1170 / -0.0239
FVC / -0.2001 / -0.0559
EU-2 / ALSFRSr / -0.1620 / -0.0231
FVC / -0.2502 / -0.0073
Table e-5: 95% confidence limits (CL) for slopes of decline for each of the comparison datasets. CL are provided for calculations at 270 days after intervention.
Trial Group / Control Group / N< Lower CL (%) / N Within CL (%) / N> Upper CL (%)
Outcome: ALSFRSr
Phase 1+2 / EU-1+2 / 2 (8) / 9 (38) / 13 (54)
Phase 2 / EU-2 / 0 (0) / 9 (60) / 6 (40)
Phase 1+2 / CEF / 10 (42) / 0 (0) / 14 (58)
Phase 2 / CEF / 8 (53) / 0 (0) / 7 (47)
Phase 1+2 / ProACT / 2 (8) / 3 (13) / 19 (79)
Phase 2 / ProACT / 2 (13) / 2 (13) / 11 (73)
Outcome: FVC % Predicted
Phase 1+2 / EU-1+2 / 4 (17) / 6 (25) / 14 (58)
Phase 2 / EU-2 / 2 (13) / 10 (67) / 3 (20)
Phase 1+2 / CEF / 6 (25) / 1 (4) / 17 (71)
Phase 2 / CEF / 3 (20) / 1 (13) / 11 (73)
Phase 1+2 / ProACT / 12 (50) / 0 (0) / 12 (50)
Phase 2 / ProACT / 9 (60) / 0 (0) / 6 (40)
Outcome: Grip Strength
Phase 1+2 / CEF / 8 (33) / 0 (0) / 16 (67)
Phase 2 / CEF / 6 (40) / 0 (0) / 9 (60)
Table e-6: Individual slope estimates from trial participants compared to the 95% confidence limits of the control datasets. Numbers in parentheses are the percentage of participants falling below, within, or above the CL for estimates at 270 days after intervention. EU=Emory control dataset. CEF=ceftriaxone placebo dataset.