Randomized clinical trial to assess the protective efficacy of a Plasmodium vivax CS synthetic vaccine

A randomized, double-blind, controlled vaccine clinical trial was conducted to assess, as the primary outcome, the safety and protective efficacy of the Plasmodium vivax circumsporozoite (CS) protein in healthy malaria-naïve (phase IIa) and semi-immune (phase IIb) volunteers. Participants (n = 35) were randomly selected from a larger group (n = 121) and further divided into naïve (n = 17) and semi-immune (n = 18) groups and were immunized at months 0, 2, and 6 with PvCS formulated in Montanide ISA-51 adjuvant or placebo (adjuvant alone). Specific antibodies and IFN-γ responses to PvCS were determined as secondary outcome; all experimental volunteers developed specific IgG and IFN-γ. Three months after the last immunization, all participants were subjected to controlled human malaria infection. All naive controls became infected and drastic parasitemia reduction, including sterile protection, developed in several experimental volunteers in phase IIa (6/11) (54%, 95% CI 0.25–0.84) and phase IIb (7/11) (64%, 95% CI 0.35–0.92). However, no difference in parasitemia was observed between the phase IIb experimental and control subgroups. In conclusion, this study demonstrates significant protection in both naïve and semi-immune volunteers, encouraging further PvCS vaccine clinical development. Trial registration number NCT 02083068. This trial was funded by Colciencias (grant 529-2009), NHLBI (grant RHL086488 A), and MVDC/CIV Foundation (grant 2014-1206).

1. The first immunization dose contained a mixture of peptides N and C only (50μg/peptide; total dose 100μg/dose), whereas for immunizations second and third, the doses comprised peptides N, R and C (50μg peptide/dose; a total of 150μg protein/dose). What is the rationale for this study design? The central repeat region has been associated with protective efficacy against other Plasmodium species including P. falciparum, P.berghei, P.yoelli. It would appear that the authors deliberately wanted to minimize he elicitation of high levels of antibodies against the repeat region.
2. There is confusion on page 13 regarding protective efficacy determined by the presence of patent microscopic parasitemia that was observed in five of 11 naïve Exp volunteers (42%) and four of 11 semi immune volunteers (36%), who did not develop parasitemia during the 60-day follow-up. This does not seem to be consistent with results in Table 2 which showed 4/11 subjects protected in the naive group versus 3/11 subjects with no parasitemia in the semi-immune group. Please provide clarity on the presentation of these results.
3. An expanded discussion is warranted about the results on reasons that lower IFN-g reactivity was observed in the semi-immune than in the naïve group.
4. There was little to no relationship between protection and levels of antibody titer by subject. As immune correlates of protection is a "hot" topic in vaccine development, whether an association between N or C terminal domains or the central repeat region correlated with risk/protection following CHMI.
5. The Methods section would benefit from providing sequences of the peptides used for immunization (L, repeat,C). 6. Figure 2 should have log rank P value shown 7. Table 2 shoed that 2 control subjects withdrew. Were these before or after CHMI? 8. Table 1 should show AEs by proportion since there there are different numbers of subjects receiving 1,2,3 doses of vaccines. There is no reference to grade of severity of AEs which should be included. 9. The authors conclude that one subject with recurrence of P. vivax malaria was due to reinfection, however there was no genotyping of these parasites to suggest that a relapse of vivax infection stemming from the CHMI may have been responsible for the 2nd infection. 10. Since subjects received different numbers of infectious mosquito bites, a table should have been included for subjects in the control and vaccine groups documenting exaqctly how many infectious mosquito bites were used. The force of infection is a determinant of liklihood of infection in CHMI.
11. The discussion would benefit from 1) how antibody or CMI (IFN-g) play a role in protection; 2) what modifications may be necessary to improve vaccine efficacy (modification to vivax sequneces or alternative adjuvants?); differences in naive subjects versus semi-immune subjects related to challenge outcome and immune responses Reviewer #2: Remarks to the Author: These are complex studies and important to conduct. Promising results are shown but there are many questions regarding the trial: -could not find the registration of this particular trial -the recruitment started in 2014, when was it completed? -when were the antibody and T cell assays performed? -the aim was to recruit 52 subjects but ended up recruiting 35, some explanation would be helpful -how were the 121 subjects asked to participate, how and where were they asked and how many were from the endemic area and how many from non endemic -then in the trial flow diagram we see 80 did not meet the inclusion criteria, which again raises the question how was the recruitment organized -the flow diagram should indicate how many from endemic and non endemic area were excluded -detail of the recruited subjects should be given in more detail and what is meant by "This study sample is not representative of the whole population of individuals living in the country's malariaendemic areas;" Points regarding results: -figures can be improved considerably -the antibodies shown are of those that react? the comparisons can be better presented -would be informative to show ELISPOT results for each individual (dot plots?) Abstract and Discussion: -not very precise when it comes to immunological reactivity -no discussion of the discrepancy between immunogenicity and protection -deeper discussion of the findings in relation to data from other studies -patterns of parasite development, the immunological data Reviewer #3: Remarks to the Author: The authors present a P. vivax CHMI vaccine study demonstrating a level of sterilizing protection in trial participants, this seems of high interest to the study of malaria vaccines. The general message of the paper is concise and clear. My main comments focus on the statistical analyses.
Major comments: I would recommend the authors have a statistician, preferably a co-author, review this manuscript to ensure that the analysis of the results is appropriate. I do not believe there are any irreparable deficiencies in the data and subsequent results, but I highlight several concerns in the statistical analysis below. 1) Issues related to protection efficacy (PE) statistics and subsequent reporting. -The PE for the semi-immune group needs to account for the infection rate in their control group (See https://doi.org/10.1093/infdis/jiaa421 as an example). The protected semi-immune control participant cannot simply be excluded from the PE estimate as an outlier. -There is no formal statistical analysis of PE. At the very least, what is the precision (e.g., 95% CI) of these estimates? (reference above also addresses this) -There appears to be two different estimates of sterilizing PE in this study. In the abstract, sterilizing PE is reported as 5/11 (note typo in Abstract and Results, should be 45%, not 42%) and 4/11 based on "no parasitemia". This result is repeated in Results section "Vaccine efficacy", but a second set of numbers are also presented "based on the survival analysis": 4/11 and 3/11. The latter is then repeated in the lead Discussion paragraph. Unless there are multiple definitions of protection (e.g., sterilizing vs. reduction in parasitemia), there should only be one estimate of PE reported. If there were multiple definitions of protection, they should be more clearly defined and easily referenced.
-Lines 202-206 (Stats Methods), description of sample size calculation is unclear. What endpoint comparison is being powered? What is the range of effect sizes? -Lines 206: "…protection efficacy was assessed at a 5% significance level and 80% power." Is this related to the sample size calculation? What comparison (ie, effect size) was assessed for 80% power? Why aren't there any formal comparisons of efficacy performed in this manuscript?
2) Comparison of immunological endpoints: it is unclear what statistical comparisons were performed and what criteria were used to determine their inclusion in the Results. All performed tests should be clearly reported so the reader has a scope of the total hypotheses being evaluated. Generally, I was confused whether p-values represented tests over time (times often not clearly stated) or between cohorts; and there were no tables providing this information.
-There are responder definitions for both assays. It is clearly stated that all vaccinated participants were seropositive via ELISA post-all vaccinations. Is this true for the ELISPOT responses? Can a similar a sentence lead those results?
3) In addition to the binary protection-derived endpoints--sterilizing protection and delay in patency-can we learn anything from the qPCR continuous readout? For example, does parasitemia look similar among infected vaccinated and control participants? Correlations between immunological and parasite endpoints are generally of interest, but perhaps that is a follow-up analysis.

Reviewer # 1 Reviewer statement
The manuscript presents the first results of a P. vivax vaccine that demonstrates protective efficacy against infection following controlled human malaria infection challenge. The following comments are intended to provide clarity to readers concerning study design and to provide rationale for several discrepancies contained in the document. QUESTION 1. The first immunization dose contained a mixture of peptides N and C only (50μg/peptide; total dose 100μg/dose), whereas for immunizations second and third, the doses comprised peptides N, R and C (50μg peptide/dose; a total of 150μg protein/dose). What is the rationale for this study design? The central repeat region has been associated with protective efficacy against other Plasmodium species including P. falciparum, P.berghei, P.yoelli. It would appear that the authors deliberately wanted to minimize he elicitation of high levels of antibodies against the repeat region.

RESPONSE 1. This concept is now included on Vaccines section, page 7, lines 118-120.
"In this immunization scheduled we deliberately wanted to diminish the levels of antibodies to the R region known to be immunodominant 28 , to induce a balanced response to the three protein fragments 17 ." The interpretation of the reviewer is correct. We wanted to balance the levels of antibodies produced against the three protein regions. The central repeats (R) domain of the CS proteins of all studied Plasmodium species is composed of tandem repeats of the same amino-acid sequence [Eg.,19 or ]. This structure confers higher immunogenicity to the central R than the amino (N) and the carboxyl (C) regions, generating an imbalance or immunodominance. While the central R domain is associated with protection, the N and C fragments contain important functional stretches such as the RI in the N-terminal region (Coppi, A et al, 2011 PMID:21262960) of the protein and the RII and neighbor sequences in the C-terminal fragment (ref 17), potentially equally crucial for protection.
In a previous phase I trial using the same immunogens, we had shown that it was possible to balance the recognition of the three protein fragments by using the vaccination regime similar to the one used here (ref 17). We believe that this immunization strategy influences the protective efficacy. Q 2. There is confusion on page 13 regarding protective efficacy determined by the presence of patent microscopic parasitemia that was observed in five of 11 naïve Exp volunteers (42%) and four of 11 semi-immune volunteers (36%), who did not develop parasitemia during the 60-day follow-up. This does not seem consistent with Table 2, which showed 4/11 subjects protected in the naive group versus 3/11 subjects with no parasitemia in the semiimmune group. Please provide clarity on the presentation of these results. The precise related changes are described below in responses to reviewers 2 and 3. Q 3. An expanded discussion is warranted about the results on reasons that lower IFN reactivity was observed in the semi-immune than in the naïve group. Q 4. There was little to no relationship between protection and levels of antibody titer by subject. As immune correlates of protection is a "hot topic" in vaccine development, whether an association between N or C terminal domains or the central repeat region correlated with risk/protection following CHMI. Here, we have found reactivity to the three different protein regions evaluated (N, R, and C). Responses to the three protein fragments were higher in the naïve (phase IIa) groups, although without correlation between antibodies to the individual fragments that could identify a correlate of protection, suggesting that antibodies to multiple critical epitopes are additive and contribute to protection. In our previous study (ref 17), the three protein fragments were shown to induce sporozoite invasion inhibition (ISI assays).
On the other hand, the correlates of protection (CoP) are indeed a "hot topic." The importance of neutralizing antibodies for protection against sporozoite invasion has been demonstrated experimentally in vitro and in vivo (ref 37-38) using monoclonal antibodies. However, protection does not seem to result from a single specificity or cytokine activity either in nature or experimentally in humans. We favor the idea of multiple antibodies to various critical epitopes, in combination with cytokines and potentially other immune effectors. System biology approaches are expected to contribute to this task. Despite recent studies providing valuable data on human naïve and semi-immune volunteers' response to the Pf-RTS,S, and Pf-R21 vaccine candidates, there is no clear understanding of CoP or immune signatures protection (ref 39-40). As mentioned above, a recent NIH-RFA precisely addresses this matter, which is common to other vaccines and micro-organisms.
Q.5. The Methods section would benefit from providing sequences of the peptides used for immunization (L, repeat, C). Figure 2, page 23 Q.6. Figure 2 should have log rank P value shown R.6. Fig 2 (now Figure 3) was edited Q. 7. Table 2 showed that 2 control subjects withdrew. Were these before or after CHMI?

R.7. Volunteers withdrew before CHMI.
It was described in the original Figure 1, page 22. It is now also mentioned in Results section, page11, lines 214-217. Table 1 should show AEs by proportion since there are different numbers of subjects receiving 1, 2, 3 doses of vaccines. There is no reference to grade of severity of AEs which should be included.
For more clarity, we have changed table1 on page 27.

All volunteers received the same number of vaccine doses (3).
Q. 9. The authors conclude that one subject with recurrence of P. vivax malaria was due to reinfection; however, there was no genotyping of these parasites to suggest that a relapse of vivax infection stemming from the CHMI may have been responsible for the 2nd infection.

R.9. The reviewer is correct. The new case developed two months after the close follow-up had ended. The volunteer was in a rural setting in the endemic area, distant from our main campus in Cali. He was diagnosed and treated at one of the points of care of the National Malaria Control Program. We were informed about it after the patient had been drug-treated and did not have the opportunity to access the parasites.
Q.10. Since subjects received different numbers of infectious mosquito bites, a table should have been included for subjects in the control and vaccine groups documenting exaqctly how many infectious mosquito bites were used. The force of infection is a determinant of liklihood of infection in CHMI.

R.10. Table 2, page 28 was modified. A column with the number of infected mosquito bites per volunteer, as well as an index of potential number of sporozoites/bite, was included. Also Methods section on page 9, lines 162.
Volunteers were exposed to 2-4 infected mosquito bites confirmed by mosquito dissection after the CHMI. We have found here and in previous CHMI that there is no difference in prepatent period or parasitemia density, and the number of infective bites, even in a range between 2 and 10 infected mosquito bites (ref 19).
Q.11. The discussion would benefit from 1) how antibody or CMI (IFN-g) play a role in protection; 2) what modifications may be necessary to improve vaccine efficacy (modification to vivax sequneces or alternative adjuvants?); differences in naive subjects versus semi-immune subjects related to challenge outcome and immune responses R.11. This has been expanded in the Discussion section, page 16, lines 323-331. We do not envisage any modification; instead, we plan to conduct a larger trial to improve the power of the studies and reduce the heterogeneity of the volunteers.

Reviewer #2
These are complex studies and important to conduct. Promising results are shown but there are many questions regarding the trial:

R.2. As described in the Study Design and Participants section, page 6, line 96.Volunteers' recruitment started on 03 October 2014 (first patient in) and was completed on 22 December 2014 (last patient in).
Q.3 -when were the antibody and T cell assays performed ?

R.3. T-cell studies were performed on the day of the volunteers' bleeding.
Q.4. -the aim was to recruit 52 subjects but ended up recruiting 35, some explanation would be helpful.

As described in the recruitment process, 121 volunteers attended the invitation to participate. We intended to identify 52 potential participants fulfilling the inclusion criteria and enroll 36. The inclusion criteria included several conditions that made enrolment challenging in an endemic population (Appendix 3). Unfortunately, 80/121
did not meet the inclusion criteria and had to be excluded. In addition, six declined their participation before enrolment; thus, we decided to start with the 35 available to avoid delays and greater losses. Q.5. -how were the 121 subjects asked to participate, how and where were they asked and how many were from the endemic area and how many from non-endemic.

R.5. This process was further explained in Section Volunteers enrolment and retention, page11, lines 205-208.
The volunteers were invited to participate by different means, including posters, flyers, radio, and TV broadcasting, followed by meetings and workshops to explain the study's objectives, risks, and benefits. However, in contrast to the other five clinical trials previously conducted by our group, this trial involved a longer duration. In addition, volunteers from the endemic region demanded to stay abroad for at least two months.
Volunteers were simultaneously recruited in Cali (phase IIa) and Buenaventura (phase IIb). Thirty-eight volunteers were initially from malaria-free areas and 83 from endemic sites (See Figure 1, Flow diagram, page 22). Q.6. -then in the trial flow diagram we see 80 did not meet the inclusion criteria, which again raises the question how was the recruitment organized.

R.6.
This question is answered to this reviewer in Responses 4 and 5 above. This strengthens the challenge of these trials in endemic communities, even with the experience of several previous clinical trials in the same region (refs 16-19, 21). On the other hand, regarding the "pattern of parasite development," the study protocol indicated immediate treatment of the infection upon confirmation in blood circulation (patent parasitemia); then, it is impossible to determine its parasite development pattern beyond treatment. In this sense, the survival curve presented in Figure # 5 corresponds to the appearance of the parasite in circulation in those volunteers who were not fully protected.

Reviewer #3 (Remarks to the Author):
The authors present a P. vivax CHMI vaccine study demonstrating a level of sterilizing protection in trial participants, this seems of high interest to the study of malaria vaccines. The general message of the paper is concise and clear. My main comments focus on the statistical analyses.

Major comments:
I would recommend the authors have a statistician, preferably a co-author, review this manuscript to ensure that the analysis of the results is appropriate. I do not believe there are any irreparable deficiencies in the data and subsequent results, but I highlight several concerns in the statistical analysis below.

Issues related to protection efficacy (PE) statistics and subsequent reporting.
Q.1-The PE for the semi-immune group needs to account for the infection rate in their control group (See https://doi.org/10.1093/infdis/jiaa421 as an example).

R.1-We defined the infection rate taking into account the protection cutoff of the Phase IIa study and the analysis is explained in lines 287-290. Specifically, on line 288, we stated that " in the control group (n=5) three volunteers were considered protected (infection rate 3/5 = 40%) based on the protection cutoff of the Phase IIa study.
To be more explicit we suggest to insert this sentence (infection rate 3/5 = 40%) on line 288.

Q.2-
The protected semi-immune control participant cannot simply be excluded from the PE estimate as an outlier.

R.2-On line 287 we have stated "
In the Ctrl group, a volunteer did not develop patent parasitemia, and two others (total 3/5) (60%, 95% CI 0.31-0.89) developed parasitemia levels below 80 parasites/mL", which indeed may continue communicating the message that the volunteer was not taking into account as control.
We now suggest the following sentence (line 289) "Using the protection cutoff of the Phase IIa study, three volunteers of the semi-immune Ctrl group were considered protected as parasitemia levels were below 80 parasites/L, including the volunteer (CSI 572) that did not develop patent parasitemia".
Q.3-There is no formal statistical analysis of PE. At the very least, what is the precision (e.g., 95% CI) of these estimates? (reference above also addresses this) R.3-We appreciate this recommendation and it was revised following the reviewer suggested publications. The 95% CI data was included in the Abstract,page 2,line 11;and in Results section,Vaccine Efficacy, Q.4 There appears to be two different estimates of sterilizing PE in this study. In the abstract, sterilizing PE is reported as 5/11 (note typo in Abstract and Results, should be 45%, not 42%) and 4/11 based on "no parasitemia". Revise crude data Q.5-This result is repeated in Results section "Vaccine efficacy", but a second set of numbers are also presented "based on the survival analysis": 4/11 and 3/11. The latter is then repeated in the lead Discussion paragraph. Unless there are multiple definitions of protection (e.g., sterilizing vs. reduction in parasitemia), there should only be one estimate of PE reported. If there were multiple definitions of protection, they should be more clearly defined and easily referenced.

Q.11
All performed tests should be clearly reported so the reader has a scope of the total hypotheses being evaluated. Generally, I was confused whether p-values represented tests over time (times often not clearly stated) or between cohorts; and there were no tables providing this information.

R.11-
We have tried to be as clear as possible to facilitate the comprehension of the study. p values were calculated between cohorts.
Q.12 -There are responder definitions for both assays. It is clearly stated that all vaccinated participants were seropositive via ELISA post-all vaccinations. Is this true for the ELISPOT responses? Can a similar a sentence lead those results?

R.12-
The recommended sentence about ELISPOT responses was included in the Results section, page 13, line 254. IFN-production was variable and at different time points there were negative Q.13 In addition to the binary protection-derived endpoints--sterilizing protection and delay in patency--can we learn anything from the qPCR continuous readout? For example, does parasitemia look similar among infected vaccinated and control participants? Correlations between immunological and parasite endpoints are generally of interest, but perhaps that is a follow-up analysis.-https://doi.org/10.1371/journal.pcbi.1005255 is a theoretical overview of parasitemia assessment.

R.13-
The qPCR was performed retrospectively. It was found that in both, control and experimental semi-immune non-protected subjects parasites were detected 2-3 days before in comparison to microscopy detection. However, in the case of naïve group no differences was observed on the day of parasite detection by the two methods.
In addition, apart of the sterily protected volunteer all others displayed highly variable parasite densities, ranging from 20-400 parasite/L by microscopy (see Table 2, page 28). We did not find correlation between antibody titers and microscopic parasitemia-frequency or density, neither at group or individual level, nor with IFN- or the association of both.

Minor comments
Q.14 -The authors clearly have a great understanding of the evolving vaccine technology, but the introduction would benefit if it had a little more clinical/epidemiology background.
R. 14-The description of P.vivax clinical manifestations and epidemiology was expanded in page 3, lines 23-33.
Q.15. How is PE assessed throughout studies, linking PE between field and CHMI studies, etc.. For example, [doi: 10.1016For example, [doi: 10. /j.pt.2016.001] appears to be a relevant review including description of the authors' past efforts.
R. 15-We find this statistical calculation and models highly valuable for malaria vaccine efficacy. However, case of P. vivax CHMI analysis deserves the analysis and report in a separate publication.