How vaccine dose and delivery method affect protein production and antibody response to a bird flu mRNA vaccine in poultry

The self-amplifying RNA (saRNA) (Figure 1) was produced from a linearized plasmid (pDNA) by in vitro transcription (IVT). The plasmids are derived from VEEV strain TC-83 and contain VEEV derived-5′ and 3′ UTRs and promotors. Additionally, substitutions in the 5′ UTR (r.3a > g) and in nsP2 (p.Q739L) are present in order to decrease cytopathogenic effects upon administration. The coding sequence of the structural proteins was replaced by either the reporter gene firefly luciferase or the head domain of the hemagglutinin (HA) antigen of influenza A H5N1 (A/Anhui/1/2005(H5N1); accession number HM172104). The construct was subsequently codon optimized for use in chickens. E. coli DH5α bacteria were transformed and subsequently cultivated in liquid broth (tryptone (10 g/l; VWR International, Leuven, Belgium), yeast extract (5 g/l; Invitrogen Merelbeke, Belgium), sodium chloride (10 g/l; VWR International, Leuven, Belgium)) containing ampicillin (0.1 mg/ml; PanReac Applichem ITW reagents, Darmstadt, Germany). Subsequently, the plasmids were isolated from these bacteria using the Plasmid Plus Midi kit (QIAGEN, Hilden, Germany) and was linearized using the I-SceI restriction enzyme (New England Biolabs, Ipswich, MA, USA). IVT was performed using the MEGAscript T7 Transcription kit (Life Technologies, Merelbeke, Belgium). SaRNA was co-transcriptionally capped with a cap1 structure (CleanCap AU (TriLink Biotechnologies, San Diego, CA, USA); 100 mM). Following IVT, the saRNA was treated with DNase I (Invitrogen, Merelbeke, Belgium) in order to remove template DNA and then purified using the Monarch RNA Cleanup kit (New England Biolabs, Ipswich, MA, USA). Additionally, cellulose purification was performed as described previously in order to remove double-stranded RNA (Zhong et al. 2021). The saRNA was stored at −80 °C. SaRNA integrity was evaluated using gel electrophoresis as described by Aranda et al. (Aranda et al. 2012). Once the quality of the saRNA was confirmed, the luciferase-encoding saRNA was formulated into LNPs by Acuitas Therapeutics (Vancouver, BC, Canada). Upon return, the functionality of said saRNALNPs- was evaluated by comparative transfection of HeLa cells against Lipofectamine Messenger Max (2:1) (Thermo Fischer Scientific, Invitrogen, Merelbeke, Belgium). The size, polydispersity index and zeta potential of the luc-saRNALNPs were determined- using a Nano ZS90 (Malvern Pananalytical, Worcestershire, UK) and were, respectively, quantified at 89 nm ± 16 nm, 0.137 ± 0.016 and −6.2 mV ± 0.5 mV (mean ± standard deviation). Calculations regarding these parameters employed a refractive index of 1.330 and a viscosity of 0.8872 mPa.s.

The HA-encoding saRNA was encapsulated in-house. LNP components included ALC-0315 (MedChemExpress; Monmouth Junction, NJ, USA) as the ionizable lipid, 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE) (Avanti Polar Lipids; Alabastar, AL, USA), 1,2-dimyristoyl-sn-glycero-3-methoxypolyethyleneglycol 2000 (DMG-PEG2000) (Avanti Polar Lipids; Alabastar, AL, USA) and cholesterol (Avanti Polar Lipids; Alabastar, AL, USA). Lipid components were dissolved in ethanol at the following molar ratio (ALC-0315: 50%; DMG-PEG2000 1.5%; cholesterol: 38.5%; DOPE: 10%) and saRNA was dissolved in NaOAc buffer (7.5 mM, pH 4.5; ThermoFischer; Merelbeke, Belgium). These two phases were thoroughly mixed together in order to allow sufficient RNA encapsulation. The mixture was dialyzed overnight at 4 °C in PBS (pH 7.4) using a Slide-A-Lyzer Dialysis Cassette 20 K MWCO (Thermo Fisher Scientific, Waltham, MA, USA). Encapsulation efficiency was determined by the Quant-it RiboGreen RNA Assay Kit (Thermo Fisher Scientific, Waltham, MA, USA) per manufacturer's instruction. The size, polydispersity index, and zeta potential of the saRNA-LNPs were determined using a Nano ZS90 (Malvern Pananalytical, Worcestershire, UK). The size and zetapotential of the in-house saRNA-LNPS were, respectively, quantified at 146.5 nm ± 4.7 nm and −3.0 mV ± 1.1 mV (mean ± standard deviation). Ribogreen assay showed 80% encapsulation efficiency. Calculations regarding these parameters employed a refractive index of 1.330 and a viscosity of 0.8872 mPa.s.

One-day-old commercial ROSS 308 broilers chicks or eighteen-day embryonated ROSS 308 broiler eggs were purchased from a local hatchery (Vervaeke-Belavi, Tielt, Belgium). The embryonated eggs were hatched using a Covina Super incubator (Aveve, Wilsele, Belgium) at 37 °C. Chicks were housed on the floor covered with wood shavings in a climate-controlled facility with ad libitum access to water and feed (Versele Laga FARM mash 1 during the first ten days and mash 2 for the remainder of the experiment (Versele-Laga, Deinze, Belgium)). The experiments were conducted with the approval of the ethical committee of the Faculty of Veterinary Medicine, Ghent University (EC 2019/32 and EC 2020/03).

In vivo experiment 1: Evaluation of potential administration routes in terms of saRNA translation and humoral immune response

One-day-old commercial ROSS 308 broiler chicks or eighteen-day embryonated ROSS 308 broiler eggs were randomly allocated to 6 groups of 6 animals each and received 1.0 µg of luciferase-encoding saRNA-LNPs (luc-saRNA-LNPs). The saRNA-LNPs were administered using four different administration routes, either intramuscular injection (IM) in the pectoral muscle, subcutaneous injection (SC) in the neck, ocular eye drops (OC) or in ovo injection (IO). Both dosage and formulation volume was fixed for all administration routes at 1.0 µg saRNA formulated in LNPs and diluted in 50 µl sterile, RNase free PBS (pH 7.4; contains 137 mM NaCl, 2.7 mM KCl, 8 mM Na₂HPO₄, and 2 mM KH₂PO₄; Invitrogen, Waltham, MA, USA). All chickens received a second administration on day 14 using their respective administration route (n = 6), except the IO group, which was boosted subcutaneously. Control groups were mock treated and received sterile PBS.

In vivo bioluminescence imaging was performed as described below at 6h, 1, 2, 3, 6, 9, 13 and 16 days post prime administration except in case of the in ovo group. In the IO group, first bioluminescence measurement was performed 6 h post-hatch, aligning the measurements with the other groups.

In order to assess the suitability of the saRNA as a vaccine platform in poultry, blood was collected from the wing vein to determine antibodies against the luciferase reporter two weeks after both the first and the second administration. Subsequently, serum was extracted from the whole blood samples by a 15 min centrifugation step (1000 g). On day 28, all chickens were euthanized by intravenous injection of sodium pentobarbital (Euthanimal 400 mg/ml; Alfa Med Ltd, Cork, Ireland). A scheme of the experiment with the time points of imaging, serum collection and euthanasia can be found in Figure 2.

In vivo experiment 2: Evaluation of dose-response and effect of a second administration of saRNA-LNPs

One-day-old ROSS 308 chicks were randomly allocated into 6 groups of 12 chickens each. Next, a dose of 0.20, 1.0 or 5.0 µg luc-saRNA-LNPs, diluted in RNase free PBS, was administered either SC or IM. At 14 days of age, six chickens from each group received a booster using the same saRNA dose and administration route as used in the prime. Mock treated chickens (n = 6), injected with sterile PBS at day 1 and day 14 of age, were included as controls. Bioluminescent imaging, serum collection and euthanasia were performed as described in in vivo experiment 1. A scheme of the experiment with the time points of imaging, serum collection and euthanasia can be found in Figure 6.

In vivo bioluminescence imaging of the chickens was performed using the IVIS Illumina III imaging system (PerkinElmer, Zaventem, Belgium). D-luciferin (PerkinElmer, Zaventem, Belgium) was administered SC at a dose of 150 mg/kg into the inguinal fold. Subsequently, anesthesia was induced at 5% isoflurane (Abbott Logistics, Breda, The Netherlands) and maintained on 2.5% isoflurane via a mask while being placed on the imaging platform. Kinetic studies revealed a bioluminescent plateau at 20 min post D-luciferin administration (data not shown). Hence, chicks were imaged 20 min post D-luciferin injection using an automated exposure time to ensure that the acquired signal was within effective detection range. Bioluminescence values were quantified by measuring photon flux in photons per second (p/s) in the region of interest where bioluminescence signal arose using the Living Image Software provided by PerkinElmer. The total amount of protein produced can be estimated by calculating the area under the curve (AUC). As luciferase catalytic activity has a half-life of 3 h in vivo (Thompson et al. 1991), its measurement reflects active translation of protein of the delivered saRNA. All reported AUC values are calculated including the last IVIS measurement at day 16, but without extrapolating the results until day 28. Signal background, determined by measuring the bioluminescent signal of control chickens treated with sterile PBS, was subtracted from the bioluminescent signal in order to be able to statistically compare to bioluminescent signal provided by each administration route. However, the graph displaying the expression data (Figure 3) shows the non-background subtracted data in order to correctly represent the difference between the research samples and the negative controls.

One-day-old commercial ROSS 308 broiler were randomly allocated into 3 groups of 8 animals each. Two of these groups received 1.0 µg H5- encoding saRNA-LNPs in 50 µl RNase free PBS (pH 7.4; contains 137 mM NaCl, 2.7 mM KCl, 8 mM Na₂HPO₄, and 2 mM KH₂PO₄; Invitrogen, Waltham, MA, USA), whereas the third group received a mock formulation consisting of 50 µl of the same sterile, RNase free PBS. One of the vaccinated groups, referred to as the prime-boost group, received a second, booster administration on day 14. The other vaccinated group is referred to as the single-shot group. The control group received a second mock injection. Blood was collected from the wing vein to determine antibody production two, four and six weeks post hatch. Subsequently, serum was extracted from the whole blood samples by a 15 min centrifugation step (1000 g). On day 42, all chickens were euthanized by intravenous injection of sodium pentobarbital (Euthanimal 400 mg/ml; Alfa Med Ltd, Cork, Ireland). A scheme of the experiment can be found in Figure 10(A).

Nunc Maxisorp^™ high protein-binding plates (ThermoFischer Scientific, Waltham, MA, USA) were coated with 10 μg/mL recombinant firefly luciferase (Promega Corporation, Madison, WI, USA) for one hour at 37 °C. Plates were washed three times with washing buffer containing 0.05% Tween 20 (Sigma Aldrich, Saint Louis, MO, USA) in PBS (Gibco, Merelbeke, Belgium) and subsequently blocked overnight in blocking buffer containing 5% skim milk powder in PBS (Merck Millipore, Burlington, MA, USA) at 4 °C. Next, 100 µl of the serially diluted samples (1/10, 1/40, 1/160, 1/640) (blocking buffer) was incubated for 1h30 at RT. Afterwards, plates were washed 5 times and goat anti-chicken IgY Fc antibodies conjugated to horseradish peroxidase (ab112821; Abcam, Cambridge, UK) diluted 1:10 000 in blocking buffer, were added. After one hour at RT, plates were washed with PBS and 3,3′,5,5′-tetramethylbenzidine (TMB) (Biolegend UK Ltd., London, UK) substrate was added and incubated for 15 min. Afterwards, the reaction was stopped with stop solution (Biolegend UK Ltd., London, UK) and absorbance was measured in a microplate reader Biochrom EZ Read 400 Microplate Reader (Cambridge, United Kingdom) at 450 and 660 nm (reference wavelength). ELISA experiments were conducted in duplicate. ELISA cut-off values were defined as the average absorbance + three times the standard deviation of the absorbances of the corresponding negative control group.

The hemagglutinin inhibition (HI) assay was performed as described by Kaufmann et al. (Kaufmann et al. 2017). First, HA titration of the inactivated A/Anhui/1/2005 (H5N1) virus (NIBSC, Herts, UK) was determined using a 0.75% turkey blood cell suspension. Next, the serum was ¼ diluted with cholera filtrate incubated overnight at 37 °C, followed by 30 min incubation at 56 °C to inactivate cholera filtrate. 25 µl PBS was administered to each well of a 96 V-bottom plate. Subsequently, the first wells of each row received 25 μL of serum mixture and 2-fold serial dilutions were performed. The lowest dilution factor of serum is 8. Next, 25 μL of the inactivated strain matched H5N1 virus containing 4 HA units was added to each well. The plate was gently tapped and incubated for 30 min at room temperature. Finally, 50 μL of 0.75% turkey blood cell suspension was added to each well and incubate for 30 min. The plate was then tilted vertically for 25 s before read out.

Statistical analyses were performed using GraphPad Prism software (version 8.4.3, GraphPad, San Diego, CA, USA). Luciferase expression was analyzed post background subtraction using the respective negative control group. Quantitative luciferase expression was determined by using the area under the curve (AUC). A Kruskal Wallis test, the non-parametric variant of one way analysis of variance (ANOVA), was conducted. Correction for multiple comparisons (Dunn's multiple comparisons test) was included. When comparing administration routes, each route was compared to one other. When comparing doses, each dose administered using IM injection was compared to each other.

Significant differences between ELISA and HI titers were determined by using a Kruskal Wallis test (nonparametric) as the conditions of normal distribution and equality of variance could not be guaranteed for each group. Correction for multiple comparisons (Dunn's multiple comparisons test) was included. When comparing administration routes, each route was compared to each other. For statistical analyses of the effect of dosage and multiple saRNA administration, the antibody titers obtained after the first administration were compared to each dose and the negative control for each administration route (IM or SC) separately. The antibody titers obtained after the second administration (booster), albeit mock or saRNA, were compared to each dose and the negative control. Additionally, each mock-boosted was compared to their respective boosted group. No inter-administration route testing occurred in the dose-response trial. All statistical tests were performed at a global 5% significant level. The data are represented as means ± standard error of the mean (sem).

Eighteen-day embryonated broiler eggs or one-day-old chicks (n = 6) were immunized with 1.0 µg luc-saRNA-LNPs using either the in ovo (IO), intramuscular (IM), subcutaneous (SC), or ocular (OC) route. All chickens, including the hatched chickens in the IO group, received a second injection on day 14 using the same delivery route as the first injection except for the IO group, which received the second injection SC (Figure 2). For each administration route, a control group was injected with PBS. As is commonly observed in broilers, two chickens in the control group died during the first week after hatch. One death could be attributed to heart failure, the other chick died due to umbilical infection (omphalitis). Mortality or disease symptoms were not encountered in any of the chickens receiving saRNA-LNPs.

Shortly after the first administration, the IM, SC and IO routes resulted in high expression that slightly dropped from day 3 onwards. In contrast, ocular delivery of saRNA-LNPs by eye drops did not induce significant expression (Figure 3).

The IM, SC and IO groups displayed similar expression kinetics. In detail, both the IO and SC routes resulted in a first expression peak 6 h (0.25 day) post hatch/prime injection, followed by a small decline and a subsequent new peak at 2 days post hatch (IO) or 3 days after the first injection (SC). After this second peak, the expression signal decreased. IM administration resulted in the highest expression peaks at 6h and 3 days post prime. However, a steeper decline of the expressed signal compared to the SC and IO group could be observed after the second peak. Thirteen days after the first administration, the bioluminescent signal in all three groups still largely exceeded the bioluminescent signal of the negative control group. After the second administration, the expression reached similar peak levels as after the first administration in all administration routes. Expression levels beyond 16 days of age could not be monitored because the chickens grew too large to fit in the imaging system.

In contrast to the kinetics, the biodistribution of the expressed protein differs between the delivery routes. Bioluminescence imaging shows that SC and IM injection of saRNA-LNPs generates a local expression pattern near the injection site, whereas IO injection results in a more widespread, diffuse luciferase expression spread over the entire body (Figure 4). At day 16 however, the IO group displayed a localized expression pattern, which can be attributed to the SC administration of the booster. Still, leftover diffuse expression remains visible at day 16. The IO administered PBS-buffered control resulted in a higher background signal than the other routes (Figure 3A). This can be attributed to the presence of white feathers, which were not removed in the IO group due to the broad expression area. To adjust for this difference, the area under the curve (AUC), which is a measure of the cumulative protein production during the whole experiment, was adjusted to the specific background of each group.

At 1.0 µg, the AUC of the IM group resulted in the highest amount of total protein produced (1.3 × 10⁹), followed by the SC (7.2 × 10⁸), IO (5.7 × 10⁸) and OC (8.2 × 10⁵) group, respectively (Figure 3B). Luciferase expression after IM, SC and IO administration was significantly different from their respective negative controls (p ≤ 0.0001 and p = 0.0193 and p = 0.0170). Additionally, the AUC was significantly different between IM and OC administration (p ≤ 0.0001), SC and OC administration (p = 0.0195), IO and OC administration (p = 0.0389) and IM and IO administration (p = 0.0325).

While these results answered the main questions for this experiment, we determined anti-luciferase antibodies as a secondary outcome. It should however be noted that administration of luciferase-encoding DNA is known to induce only low titers of anti-luciferase antibodies in mice, indicating that luciferase has only limited antigenic properties (Vandermeulen et al. 2007; 2009; Petkov et al. 2013). Keeping this in mind, serum samples of all chickens were collected two and four weeks post hatch. The first samples allow to estimate the efficacy of a single shot vaccination, whereas the serum samples collected on day 28 demonstrate the effect of a booster vaccination. Despite the low immunogenicity of luciferase, seroconversion could be observed in some animals after prime vaccination (IM: 17%, SC: 67%, OC: 0%, IO: 83%) (Figure 5). The corresponding titers were, respectively, 107, 147, 0 and 348. The difference between the IO group and both the OC and the PBS group turned out to be statistically significant (both p-values = 0.0253).

After the second administration serum titers were unexpectedly lower than after the prime (IM: 113; SC: 87; OC: 27; IO: 27), while seroconversion rates increased in the IM group, remained the same in the SC group and decreased in the IO group (IM: 83%; SC: 67%; OC: 17%; IO:17%). Only the antibody titers in the IM group turned out to be significantly higher (p = 0.0398) than the titers in the negative control group, where no anti-luciferase antibodies were detected.

Next, the effect of saRNA dosage was evaluated. Based on the previous experiment, the IM and SC administration route were selected for the dose-titration experiment using 0.20, 1.0, or 5.0 µg of luc-saRNA-LNPs. On day 0, the chickens (12 per group) received their first saRNA-LNP administration. On day 14, serum samples were collected and half of the chickens (n = 6) received a second injection. Serum samples of all chickens were collected again on day 28. Bioluminescence was measured 6h, 1, 2, 3, 6, 9, 13, and 16 days post prime injection. One chicken (probably a non-starter) died two days post saRNA-LNP injection after receiving a single 0.20 µg dose. Besides cachexia and anorexia, significant lesions were not observed after postmortem evaluation.

Each of the administered doses resulted in a clear luciferase expression with a kinetic profile similar to that observed during the first in vivo trial (Figure 7A). A dose of 0.20 µg saRNA-LNPs resulted in clear protein production with an AUC of 5.30 × 10⁷ (Figure 7B). Increasing the dose to of 1.0 and 5.0 µg resulted in statistically significant higher protein production (AUC, respectively, 2.98 × 10⁸ and 2.95 × 10^8, respective p-values: 0.0001 and 0.0004). Still, the additional increase is only 4-fold, which is somewhat surprising as the dose is increased 5-fold (1.0 µg;) or even 25-fold (5.0 µg).

Two weeks after the initial administration, the mean anti-luciferase antibody titers of the single-shot group were 110, 213, and 108 in, respectively, the 0.20, 1.0, and 5.0 µg dose group. The group meant to test the prime-boost setting resulted in titers of 213, 113, and 33 in, respectively, the 0.20, 1.0, and 5.0 µg dose group. No statistical difference between doses could be discerned after prime injection. Respective seroconversion rates were 50%, 33%, and 33% in the single-shot setting and 33%, 33%, and 33% in prime-boost setting (prior to boosting).

Subsequently, half of the chickens (n = 6) received a second, booster injection. Results of the anti-luciferase IgY antibody ELISA and seroconversion rates are visualized in Figure 8. Two weeks after the second administration, the mean anti-luciferase antibody titers were 560, 45, and 275 in, respectively, the 0.20, 1.0, and 5.0 µg boosted groups and 428, 218, and 220 in the single-shot groups which indicates that antibody titers in the single-shot vaccinated groups do not drop after two weeks but continue to increase. All boosted groups and the 5.0 µg single-shot group achieved 100% seroconversion, whereas the single-shot 0.20 and 1.0 µg groups showed an increase in seroconversion to 83% at day 28. Boosted groups generally showed both higher antibody titers and higher seroconversion rates compared to mock-treated groups, but it is noteworthy that the boosted 0.20 µg group resulted in the highest mean antibody titers. Statistical significant differences between the negative control group and 3 treated groups could be observed: the boosted 0.20 µg group (p-value = 0.0003), the single-shot vaccinated 0.20 µg group (p-value = 0.0003) and the boosted 5.0 µg group (p-value = 0.0209).

When administering the saRNA-LNP formulation subcutaneously, a dose of 0.20 µg resulted in the highest amount of total protein produced (8.4 × 10⁷), followed closely by a dosage of 1.0 µg (8.3 × 10⁷) (Figure 7C,D). A dose of 5.0 µg resulted in the lowest amount of total protein produced (6.2 × 10⁷). None of the doses differed significantly from each other, yet all of them differed significantly from the negative control (5.0 µg: p = 0.0429; 1.0 µg: p = 0.0015; 0.20 µg: p = 0.0005).

Two weeks after the first injection, the mean anti-luciferase antibody titers were 17, 140, and 2 in, respectively, the single-shot 0.20, 1.0, and 5.0 µg dosage group (Figure 9). The prime-boost group resulted in titers of 115, 60, and 2, respectively. The corresponding seroconversion rates in the single-shot group were 67%, 50%, and 50% while seroconversion rates in the prime-boost group were 50%, 50%, and 17%. Half of the chickens in each group were boosted at day 14. Four weeks after the single-shot vaccination, mean antibody titers were higher than after two weeks (0.20 µg: 348; 1.0 µg: 215; 5.0 µg: 113) and the seroconversion rates were 67%, 50%, and 33% in, respectively, the 0.20 µg, 1.0 µg and or 5.0 µg dosage group. Two weeks after the boost, mean titers were higher than the ones reported after a single-shot vaccination and equaled 453, 427, and 163, respectively, in the 0.20 µg, 1.0 µg and or 5.0 µg dosage group. Seroconversion rates also increased after the booster were 83%, 67%, and 83% in the 0.20, 1.0, and 5.0 µg group. After the boost, antibody titers were not significantly different between dosages, however, the 0.20 µg group showed borderline significant higher titers (p = 0.0510).

To generate a proof-of-concept with an immunodominant antigen, saRNA encoding the secreted head domain of the HA antigen (HA-HD) of H5N1 avian influenza was produced. Chickens were IM immunized with 1.0 µg of H5-encoding saRNA encapsulated in in-house formulated LNPs. On the day of hatch, the chicks (n = 8) received the prime saRNA-LNP administration. Two weeks later (on day 14), serum samples were collected and one group, referred to as the boosted group, received a second injection. The second group, referred to as the single-shot group, received only the prime injection and was instead injected with a mock formulation on day 14. An additional negative control group received PBS at all moments of vaccination. Serum samples of all chickens were collected on day 28 and on day 42. No animals died during the course of this experiment. During statistical analysis, one outlier in the boosted group had to be removed from the data set as it deviated more than 3 times the standard deviation from the statistical mean. Experimental design is visualized in Figure 10A.

The mean HI titer of the single-shot shot group equaled 44, 60, and 44 at, respectively, 2, 4 and 6 weeks after vaccination, signaling both possible protection against mortality (titer > 32) and prevention of viral shedding (titer > 40) (Swayne and Sims 2021; Terrestrial Animal Health Code 2023; WOAH Terrestrial Manual 2025 2024) (Figure 10(B)). In the boosted group, the mean HI titer equaled 90, 320 and 264, respectively, 2, 4 and 6 weeks after vaccination (Figure 10(C)). The difference in mean HI titer between the negative control and single-shot vaccination group at 2 weeks was clear, yet is not statistically significant (p = 0.1206). Comparison between the negative control and the single-shot group at 4 weeks and 6 weeks led to the respective p-values of 0.0029 and 0.0317. When comparing the negative control and the boosted group at 4 weeks (p = 0.0026) and 6 weeks (p = 0.0077), the difference was statistically significant. The difference in HI titer between the single-shot group and the boosted group at 4 weeks and 6 weeks was significant (4 weeks: p = 0.0074; 6 weeks: p = 0.0183).