An Inconvenient Truth: Transdermal Buffering Lotions Appear to Offer No Significant Performance Improvement

Abstract

Background/Objectives: Transdermal bicarbonate (TBC) or carnosine (TC) have been sold as a convenient ergogenic aid, though little evidence supports these claims. The purpose of this study was to investigate if TBC or TC would improve high-intensity endurance cycling. Methods: Data were collected remotely using Zwift online platform. Fifteen cyclists completed four trials comprising five 9.1 km laps: warm-up lap, three interval laps (3 × 900 m hills, and 3 × 300 m sprints), and a 9.1 km time trial (TT) lap. A familiarization (FAM) trial followed by three randomized trials using a TBC, TC, or placebo (PLAC) lotion were completed. Trial data were assessed using general linear models to compare differences between conditions across cycling trials (p < 0.05). Results: Mean ± SD. 60 min mean max power (MMP60) was 2.9 ± 0.9 W/kg and ranged from 1.3 to 4.1 W/kg for participants. Exercise trials were 89.8 ± 17.0 min long. Laps 2–4 were ridden at 86.4 ± 7.3% and hill climbs at 131.6 ± 21.1% of MMP60, while sprints averaged 83.2 ± 17.7% of 30 s mean max power (MMP30s) and the TT ridden at 95.4 ± 8.7% of 20 min mean max power (MMP20). FAM trials were significantly lower (p < 0.05) in all power output measures, except TC sprints, and no heart rate or rating of perceived exertion differences. Similarly, there were no statistical differences in performance between any condition trials or placebo trial, but TBC hill climb power was significantly higher (p = 0.038) than TC trials. Conclusions: There are no apparent ergogenic benefits from TBC or TC during high intensity cycling performance.

1. Introduction

The sports supplement industry was valued at $45.2 billion in 2023 and is expected to grow 7.5% from 2023 to 2030 [1]; however, regulation of most supplements is relatively nonexistent and efficacy is often dubious, with the burden of proof placed on the consumer [2]. This makes empirical testing of manufacturer claims essential. While standard practice in the past has been to rely on “gold standard” exercise laboratory testing, such restrictions limit evaluation due to the resources required and the challenges of recruiting suitable participants to visit the lab multiple times. For most sports supplements the most important reference standard is exercise performance, which is predictably reliable day-to-day [3,4]. Recent advances in technology have given affordable lab quality tools to everyday athletes; two of the more notable tools include bicycle power meters [5] and smart bicycle trainers [6,7,8]. The advent of these tools paired with mixed reality (MR) training platforms, like Zwift, transformed indoor leisure, fitness, and competitive cycling in the last decade [9,10,11,12]. The combination of technologies now allows for remote exercise performance studies to be conducted across the world with a wide range of athletic abilities. As such, it is now possible to perform sport-related evaluations, like testing sports nutrition products, remotely.

Competitive road (indoor/outdoor) cycling relies on a complex mixture of endurance and power determinants, as well as resilience to maintain optimal performance levels near the end of a race, which has been termed “durability” [13,14,15,16,17]. It has been shown that cyclists with higher durability are more likely to be successful [15,17]. Both amateur and professional cyclists perform a variety of training sessions to improve overall performance and consequently, durability. Therefore, any ergogenic aid that can reduce fatigue could improve one’s final TT durability.

Previous work by Stellingwurf et al. [14] suggested evidence-based approaches to utilizing specific supplements to aid performance. One common target is “lactic acid”, which nearly instantaneously dissociates into H⁺ and La⁻ (lactate), because of the outdated notion that lactate causes fatigue. While lactate has been exonerated as a cause of fatigue [18,19], evidence suggests that increases in H⁺, which decreases pH, may negatively impact several aspects of muscle power output [19,20]. Therefore, supplements employing buffering agents like bicarbonate (BC) and carnosine, or its precursor beta-alanine (BA), have often been utilized in sport and exercise settings.

The preponderance of evidence indicates that both BC and BA can improve performance in very high-intensity continuous or intermittent exercise of durations from 30 s to 10 min [21,22,23,24]. More recent evidence, however, indicates that certain events lasting longer than 10 min may also benefit [22,23]. The evidence for longer duration activity seems contradictory but likely relates specifically to sports like cycling that include both intermittent and continuous high-intensity bouts of up to 10 min; examples include repeated sprints, short, steep climbs, or a final burst of power in the final 5 km of the race. Taken together, buffering supplements would theoretically improve repeated high intensity efforts or improve power in the finale (i.e., durability). To date, however, the preponderance of the evidence has been with ingested BC and BA; nevertheless, topical lotions have also been developed and marketed to athletes for at least a decade and claim similar benefits without commonly reported side effects of either buffer type [18,25].

A thorough review of the literature indicates a paucity of evidence that transdermal BC (TBC) can increase blood pH when applied at twice the dose described on the product bottle (20 g) [26], while another study using thoroughbred racehorses indicated that transdermal carnosine (TC) increased muscle carnosine levels by 46% 60 min after application [27], though no dose was reported. Despite some evidence of absorption, there is no meaningful evidence that either lotion improves exercise performance. For example, Gurton and colleagues published two somewhat conflicting studies on team sport athletes (e.g., hockey, soccer, basketball) showing that oral sodium bicarbonate (OBC) ingestion and TBC results in small, but significant improvement in repeated high-intensity performance [28], though only the OBC resulted in increased pH. The authors speculated that the interaction between the lotions menthol and sodium may have altered effort perception, but this has yet to be tested. However, a similar follow-up study in soccer players failed to replicate these findings, only showing improvement from OBC [29]. In contrast, Seah [30] failed to show any benefit from TBC to intermittent high-intensity sprints in team sport athletes. Finally, both McKay et al. [31] and Kern et al. [32] examined both changes in the acid-base balance and cycling performance in trained cyclists who performed three 30 s maximal sprints. Like prior research, no changes were seen in any of variables following application of the TBC lotion, though Kern did demonstrate possible increased blood pH.

Research results on TC are no less equivocal. Initial research by Sharpe and Macias [33], two of the three co-inventors listed on the 2017 patent filing, suggested that TC significantly improved repeated sprint and 1000 m running performance in elite male soccer players. However, the study used sequential, unblinded sessions, where the TC trials were performed last, possibly resulting in an ordering effect. Additionally, the authors appear to report clinical and statistical significance synonymously without explanation or context and the limited methodological details coupled with confusing findings make it difficult to draw meaningful conclusions from this study. A follow-up study by Harnish and Miller [34] using a randomized controlled cross-over design recruited well-trained cyclists to perform three sessions of five maximal 30 s Wingate sprints. Here the study failed to show any difference between trials in either performance, rating of perceived exertion (RPE; 1–10 scale), blood lactate concentration. More recent data from an unpublished master’s thesis [35] also failed to provide substantial evidence that TC improves repeated sprint performance in rugby players. Most notably, only three of the twelve 6 s sprints showed significantly higher peak power than the placebo (PLAC) group and there were no differences in average sprint power.

As noted at the outset, sports nutrition supplements are ubiquitous and rely on more marketing hype than science-based evidence. As we have noted here, transdermal buffering lotions would offer a convenient, easy, and cost-effective way to improve performance if they worked as marketed. However, few studies have specifically tested products as marketed, and under sports specific performance challenges. Therefore, the purposes of this study were to use a randomized controlled cross-over trial to investigate whether either TBC or TC, when applied as instructed by company recommendations, would improve either high-intensity intermittent cycling or final time trial performance during a ~50 km cycling trials conducted remotely using the Zwift MR cycling platform. Based on the available research obtained in an ad hoc literature review, we hypothesized that neither TBC nor TC would elicit a significant improvement in performance, heart rate (HR), or RPE differences beyond a similar PLAC lotion.

2. Materials and Methods

2.1. Ethics Approval and Participant Characteristics

All methodology was reviewed and approved by the Mary Baldwin University (Staunton, VA, USA) Institutional Review Board (IRB). All participants were self-reported trained cyclists recruited between May 2023 and April 2025 from across the continental USA via social media and met the following inclusion criteria: apparently healthy men and women between 18 and 55 years of age; self-reported weekly bicycle training experience such that they could cycle continuously for over 2 h; and had access to both a high-quality smart trainer and the Zwift virtual cycling platform (Zwift, Inc., Long Beach, CA, USA). However, individuals meeting all other criteria without access to Zwift were offered free access two weeks prior beginning their trials through completion of the study. Exclusion Criteria included: individuals outside the inclusionary age range; and those self-reporting any known medical condition that would preclude participation (e.g., acute musculoskeletal injuries, illness, etc.). All participants were informed of the purposes and requirements of the study and provided written consent prior to enrollment. The general study design, presentation, and results utilized the 2025 BMJ CONSORT guidelines for reporting randomized trials [36].

Once enrolled, participants completed an intake survey which provided the research team with basic physical and demographic information, endurance training characteristics, and details of recent cycling performance (see Supplement S1). They were then provided with a unique ID number used for all trial records and study files [37]. Within the context of this study, MMP values were derived from participants’ self-reported highest sustained power output for 60 min, or functional threshold power (FTP), across a 4-wk period prior to testing; where MMP60 was considered equal to FTP [37,38]. If participants were unable to provide this information, then the highest recorded power during the FAM trial was used to estimate this value. Additionally, participants were stratified into the original Zwift Categories based on individual functional threshold power (FTP) output defined as: Zwift A ≥ 4.0 W/kg; 3.9 W/kg ≥ Zwift B ≥ 3.2 W/kg; 3.1 W/kg ≥ Zwift C ≥ 2.5 W/kg; and Zwift D < 2.5 W/kg [38]. Participants’ MHR was similarly defined as the highest self-reported HR during exercise within the 3 months preceding enrollment or as the highest recorded measure across all trials in the study.

2.2. Study Design

During development of the project, it was decided that in addition to statistical significance, a performance improvement of 3–5% was likely necessary to demonstrate the efficacy of any one of the lotions provided (i.e., clinical significance) [39]. This improvement threshold was chosen based on prior research [34], as well as to accommodate 3 factors: potential day-to-day variation in performance [3]; potential placebo/nocebo effects [34,39]; and typical reported error for bicycle power meters and smart trainers [5]. A sample size calculation was conducted using this efficacy threshold in JMP Pro (version 17.0; SAS Institute Inc., Cary, NC, USA) which indicated that at least 14 participants were needed to achieve statistical power (β = 0.80; α = 0.05). This assumed that repeated measures analyses of variance (RM-ANOVA; within-factors) would be used to analyze cycling performance changes using power output data.

This study employed a randomized (Research Randomizer, www.randomizer.org) single-blind placebo-controlled 3 × 3 block crossover design with enrollment of no more than 5 participants at a time. Each participant completed 4 trials including an initial FAM trial followed by the three counterbalanced lotion trials using a TBC supplement, a TC supplement, and a PLAC. While a single researcher (CRH) was aware of which condition was provided to each subject, all trials were completed remotely from across the continental United States, so no direct interaction with participants occurred regarding lotions. All trials were conducted within the same 9.1 km MR Zwift course called “Hilly Route” (see Figure 1), where a total of 5 laps were completed for each trial.

2.3. Trial Details

Participants were provided detailed instructions on how each trial should be completed as part of the consent process and with a one-page “Quick Start” sheet that specifically outlined the order randomized-blinded lotions trials following the FAM trial. Participants were advised to maintain their usual diet and to complete all trials within the same time of day (±2 h) [40]. All trials were conducted remotely indoors within temperature-controlled settings. Participants were instructed to use at least 1 high-powered fan during their ride, drink water ad libitum, and consume at least 30–60 g of carbohydrates during each trial. Participants were asked to maintain this pattern across all 4 trials. The FAM trial was always completed first to minimize any performance or learning improvements [41,42,43,44]. Following the first trial (always a FAM trial), data were reviewed for any irregularities or deviations from trial instructions, with feedback or recommendations being communicated to participants at this time only (no further feedback was provided following subsequent trials).

As noted above and in Figure 1, each trial consisted of 5 laps along a 9.1 km virtual loop that included one 900 m hill climb and one 300 m flat sprint; these segments had distinct start and finish banners. Participants were instructed to complete lap 1 at a self-determined “easy pace” as a warm-up. Upon completing lap 1 (noted by start and finish banners) they were asked to maintain about 70% of their MMP20 for laps 2, 3, and 4, except for the hill climb and sprint segments; during which participants were instructed to “race” as fast as possible. For the final lap (lap 5), participants completed a maximal time trial (TT) with explicit instructions to evenly pace themselves across the entire 9.1 km course (i.e., not race segments); fixed duration and distance TT’s have been shown to produce highly reliable performance outcomes [45,46,47], even on virtual platforms like Zwift [48].

Each participant provided their own bicycle attached to their own smart trainer as well as their own HR monitor. Power output and HR data were recorded continuously through the Zwift app via a Bluetooth low energy connection and subsequently uploaded using trial training logs (see Section 4.4). Participants were also instructed to record their RPE (1–10 scale) following each lap and overall for each trial at least 30 min after each trial using provided trial training logs (see Section 4.4 and Supplement S2) [49]. Each trial session file was uploaded using the trial training log form and then subsequently transferred to a private Strava account (Strava Inc., San Francisco, CA, USA) where data for each lap and segment were extracted for later analysis.

2.4. Trial Training Logs

Each participant was provided a link to a Google form that could be easily completed during and/or after each trial and included a final section for Zwift file (.fit) uploads (see Supplement S2). Each trial training log included the following data points to be collected:

• Supplement application and Trial start times.
• Trial designation (FAM, X, Y, or Z).
• Supplement rating and comments, and belief it was a supplement or PLAC.
• Environmental temperature as well as pre- and post-trial body weight.
• The Zwift virtual bike and wheels used (kept unchanged across trials as they affect game speeds).
• RPE (1–10) for each lap and overall trial RPE [49].
• The volume of fluids consumed and details on carbohydrate intake.

2.5. Supplmenent and Placebo Lotions

The supplement lotions used in this study were readily available on the market that included a TBC lotion containing 0.5% menthol and 33.2% HCO₃⁻ (AMP PR, Momentous, Park City, UT, USA) and TC gel containing 1.25% menthol and 1.5% carnosine-complex (LactiGo™, Outplay Inc., Las Vegas, NV, USA). Because the appearance, consistency, smell, and menthol content of the two supplements was completely different, a popular and easy to apply lotion containing 0.8% menthol (Muscle Moisturizing Shea Butter Lotion, The Village Co. LLC, Eden Prairie, MN, USA) was used for the PLAC. Unlike prior research, our purpose was to test market claims; thus, samples of each supplement were provided to participants in accordance with single-dose packaging available for purchase at the time of testing for both TBC and TC. Precisely measured using an Ozeri ZK14-T digital scale (Ozeri Corp, San Diego, CA, USA), participants received 20 g of TBC (containing ~6.64 g of HCO₃⁻), 15 g of TC (containing ~0.15 g of carnosine), and 15 g of the PLAC lotion (PLAC dosage was selected to provide participants with similar total lotion volumes between all lotion conditions). Each lotion was shipped to participants in clear, BPA-free plastic jars marked only with supplement letter codes X (TBC), Y (TC), or Z (PLAC). Blinded to which lotion they were to apply each trial; participants were instructed to use the entire contents of each container (scraping the interior of the container) to their legs and gluteus maximus no less than 1 h prior to their expected trial start time. The application time and trial start times were recorded in the trial training log.

To assess how participant perceptions of each lotion may have influenced performance, they were asked to answer a series of questions about each lotion used in their trial logs (excluding the familiarization trial, trial 1; See Supplement S2). Specifically, they were asked to comment on the lotion (e.g., like/dislike, the smell, texture, etc.), to rate the lotion between 1 and 5 stars (1 star: Bad; 5 stars: Great), and finally they were asked whether they believed they received the supplement or PLAC. Accurate supplement identification was only considered if the participant was able to identify all lotions accurately.

2.6. Performance Measures

Performance data across laps 2–5 used for analysis included entire lap average power (laps 2–4), hill climb segment and sprint segment average power (laps 2–4), TT average power (lap 5), entire lap average HR and RPE (laps 2–4), and TT average HR and RPE (lap 5). Data from laps 2–4 were also averaged and compared to the three MMP values reported in

Table 1. Summary of baseline participant physical, performance, and training characteristics.

	Overall	Zwift A	Zwift B	Zwift C	Zwift D
Physical Characteristics
n	15	4	3	4	4
Age (years)	45.1 ± 12.1	42.0 (28.0, 50.0)	54.0 (47, 61)	46.0 (30.0, 51.0)	53.5 (18, 55)
Height (cm)	175.6 ± 6.9	176.7 (175.0, 180.3)	181.6 (177.8, 183)	170.2 (159, 182.9)	176.1 (167.6, 180.3)
Weight (kg)	74.4 ± 11.6	70.5 (65.4, 77.2)	79.8 (69.5, 88.6)	75.4 (60.5, 94.0)	76.1 (55.5, 82.0)
RHR (bpm)	51.3 ± 7.8	47.5 (45.0, 51.0)	45.0 (39.0, 56.0)	53.5 (48.0, 58.0)	58.5 (43.0, 66.0)
MHR (bpm)	177.1 ± 11.0	185.0 (185, 190.0)	172.0 (168.0, 180.0)	183.0 (161.0, 194.0)	164.5 (161.0, 181.0)
Performance Characteristics
MMP30s (W/kg)	7.1 ± 2.3	9.8 (9.2, 10.4) # D	6.8 (6.7, 6.9)	6.7 (3.6, 9.3)	5.8 (2.7, 7.4)
MMP20 (W/kg)	3.2 ± 1.0	4.4 (4.0, 4.6) * D	3.5 (3.5, 3.7)	2.8 (2.6, 2.9)	2.2 (1.5, 2.5)
MMP60 (W/kg)	2.9 ± 0.9	4.0 (3.6, 4.1) * D	3.3 (3.2, 3.4)	2.6 (2.3, 3.1)	1.9 (1.3, 2.2)
Training Characteristics
Years training (years)	13.4 ± 11.8	18.5 (3.0, 36.0)	17.0 (12.0, 31.0)	6.0 (1.0, 10.0)	7.0 (2.0, 23.0)
6-wk training Volume (min)	454.0 ± 209.7	630.0 (180.0, 300.0)	480.0 (480.0, 810.0)	315.0 (180.0, 480.0)	330.0 (90.0, 600.0)
Typical training ride (min)	123.0 ± 69.5	135.0 (90.0, 240.0)	135.0 (120.0, 240.0)	90.0 (60.0, 150.0)	75.0 (60.0, 90.0)
Typical longest ride (min)	187.3 ± 68.6	240.0 (180.0, 300.0) # D	240.0 (120.0, 300.0)	150.0 (120.0, 270.0)	150.0 (60.0, 180.0)
Typical Race Duration (min)	140 ± 86.8	60–600	60–300	60–120	60–180

Mean ± SD or Median (Min, Max). Bold denotes significant difference. # Significantly different from noted category (p < 0.05; e.g., #A). * Significantly different from noted category (p < 0.01; e.g., * B).

. Specifically, hill climb segment average power was compared to MMP60 (i.e., FTP) and sprint segment average power was compared to MMP30s (i.e., Wingate).

2.7. Statistical Analyses

Data were screened and cleaned for accuracy and completeness prior to performing any statistical analyses. Descriptive statistics were computed as means ± SD for all physical characteristics and performance measures, unless otherwise noted. Due to small sample sizes within individual Zwift category classes, Kruskal–Wallis analysis was completed followed by a post hoc Dunn test for multiple comparisons, when significance was detected. One of the main outcomes to be tested was how TBC and TC versus PLAC would affect repeated hill climb power output and sprint power output, lap HR, and lap RPE for laps 2–4. Therefore, 4 × 3 repeated measure ANOVAs and partial eta-squared effect sizes were used to compare differences in each dependent variable (power, HR, RPE) by trial condition (FAM, TBC, TC, and PLAC) and lap (2–4) over time. A second main outcome to be tested was the effect of each of the four conditions on final TT performance, which was tested using 4 × 1 ANOVAs comparing average lap power, HR, and RPE. Mauchley’s Test of Sphericity confirmed equality the variance differences between all combinations of related groups. If violated, Greenhouse-Geisser adjustments were used [50]. Follow up pairwise comparisons and simple effects tests were planned using Tukey’s LSD for all significant main effects and interactions. Finally, analysis of potential placebo effects was performed using Fisher’s Exact test for categorical variables due to the small sample sizes within survey categories. Additional analysis comparing placebo effects of those who “believed” they used a real supplement versus those who did not was performed using paired T-tests. All statistical procedures were performed with the Statistical Package for Social Sciences (version 30; IBM Corp., Armonk, NY, USA) with α < 0.05 set a priori.

Post hoc power analyses were conducted using Kendall’s W and partial eta-squares. Each effect size, alpha, and sample size for each measure demonstrated high statistical power > 0.80 for all non-parametric and parametric procedures. Finally, supplementary analyses of difference scores were examined for any possible significant differences.

3. Results

3.1. Baseline Participant Physical, Performance, and Training Characteristics

The general study flow is depicted in Figure 2. While 22 individuals met study inclusion criteria, 7 were ultimately excluded from participation, 2 individuals failed to consent, and 5 individuals consented but failed to initiate testing. A resulting total of 14 men and 1 woman (n = 15) enrolled and completed the study. Non-parametric statistical analysis indicated that Zwift Category A riders exhibited a significantly higher 30 s mean maximal power (MMP30s), as well as 20 and 60 min MMP (MMP20 and MMP60, respectively) than category D riders.

3.2. Overall Trial Characteristics and Outcomes

All participants completed testing without any adverse events reported; however, individual trial data (i.e., Zwift files) for 2 participants were lost due to technical issues. In addition, 1 of these participants also failed to complete their self-report trial training log resulting in additional data loss. Both participants declined to repeat these trials resulting in the overall loss of all data associated with 1 TBC trial as well as 1 PLAC trial and training log. Two additional participants HR data were lost due to technical failure for their familiarization trial (FAM) only (1 participant: laps 1–5; 1 participant: laps 2–5). Therefore, the n associated with each outcome measure (power, HR, and RPE) varied slightly between trials and laps, ranging from n = 14–15 among power and RPE data and n = 13–15 among HR data. All analyses were still carried out following an intention to treat paradigm. Post hoc power analyses using Kendall’s W and partial eta-squares indicated statistical power was maintained despite the loss of some data. The weighted average power across all variables was 0.897 for non-parametric effects and 0.854 for parametric effects, suggesting that our findings had sufficient sensitivity to detect moderate-to-strong effects and a relatively low probability of Type II error. Our overall findings are further supported by an analysis of trial data difference scores (Supplement S3). Trial-by-lap data are summarized in

Table 2. Performance results by lap and trial condition.

Lap	Trial Condition	Entire Lap Average Power (W)	Average HR (bpm)	RPE (1–10)	Hill Climb Segment Average Power (W)	Sprint Segment Average Power (W)
	FAM	140.1 ± 38.9	114.4 ± 10.8	3.3 ± 1.7	159.3 ± 29.3	151.2 ± 55.7
1	TBC	134.8 ± 41.3	115.6 ± 15.1	2.8 ± 1.6	151.7 ± 34.7	138.4 ± 56.8
	TC	136.3 ± 40.0	116.1 ± 15.4	3.1 ± 2.0	156.5 ± 34.0	121.3 ± 52.0
	PLAC	136.2 ± 39.2	114.5 ± 11.9	3.2 ± 1.5	153.3 ± 35.1	141.3 ± 49.6
	FAM	180.5 ± 56.3	138.4 ± 13.9	5.3 ± 1.4	267.2 ± 97.5 *	431.0 ± 211.2
AVG 2–4	TBC	180.8 ± 56.4	140.8 ± 16.1	5.3 ± 1.5	300.6 ± 101.6	483.0 ± 211.5 #
	TC	185.1 ± 56.2	141.5 ± 16.9	5.5 ± 1.6	292.9 ± 103.0	464.9 ± 207.0
	PLAC	192.8 ± 52.2	143.4 ± 15.6	5.3 ± 1.5	298.6 ± 101.2	488.5 ± 221.8 #
	FAM	181.0 ± 58.1	133.5 ± 14.6	5.6 ± 1.3	274.8 ± 100.8	437.4 ± 215.7
2	TBC	185.3 ± 57.2	137.8 ± 18.0	5.7 ± 1.6	309.4 ± 103.5	502.8 ± 222.6
	TC	187.3 ± 55.6	136.7 ± 17.3	5.9 ± 1.6	298.8 ± 111.6	468.2 ± 235.6
	PLAC	196.6 ± 52.4	138.0 ± 16.8	5.6 ± 1.6	304.6 ± 102.8	494.9 ± 214.8
	FAM	182.8 ± 55.2	136.6 ± 14.4	5.6 ± 1.5	256.0 ± 101.1	414.6 ± 223.0
3	TBC	181.3 ± 55.2	139.8 ± 17.8	6.0 ± 1.6	292.2 ± 110.0	464.4 ± 230.0
	TC	185.3 ± 54.6	140.1 ± 18.7	5.9 ± 1.6	273.0 ± 103.4	458.7 ± 225.2
	PLAC	191.6 ± 52.5	141.2 ± 16.6	5.8 ± 1.8	291.5 ± 107.8	475.5 ± 251.6
	FAM	177.7 ± 57.6	138.5 ± 12.7	6.6 ± 1.1	270.6 ± 104.2	440.9 ± 216.9
4	TBC	175.9 ± 58.8	139.5 ± 16.3	6.7 ± 1.3	300.2 ± 101.8	481.8 ± 203.9
	TC	182.8 ± 60.1	141.4 ± 15.8	6.9 ± 1.0	297.0 ± 102.1	467.7 ± 188.7
	PLAC	190.3 ± 52.9	143.6 ± 15.0	6.5 ± 1.3	299.8 ± 102.5	495.1 ± 210.8
	FAM	220.8 ± 76.8 *	152.1 ± 12.2	8.2 ± 0.9	235.2 ± 66.9	247.0 ± 74.4
5	TBC	232.9 ± 69.5	155.7 ± 13.6	8.5 ± 1.2	246.3 ± 69.9	240.7 ± 56.9
	TC	229.5 ± 67.3	155.4 ± 15.4	8.6 ± 1.2	243.4 ± 65.2	244.7 ± 63.6
	PLAC	233.0 ± 73.1	156.8 ± 14.0	8.3 ± 1.1	245.3 ± 72.9	236.3 ± 69.8

Mean ± SD. Bold denotes significant difference. * Significantly different from all other trials within row (p < 0.001). # Significantly different from FAM trial (p < 0.01). Lap 1 data not analyzed.

.

3.2.1. Trial Characteristics

Neither environmental temperature (67.6 ± 5.4 °F; 1 missing report: 1 PLAC trial) nor trial weight loss (−0.6 ± 0.6 kg; 2 missing reports: 1 FAM trial, 1 PLAC trial) differed significantly across trials (respectively, p = 0.969 and p = 0.941. Participants also ingested similar amounts of fluid (0.9 ± 0.5 L) and carbohydrates (40.8 ± 38.0 g; across all trials (respectively, p = 0.849 and p = 0.879). Overall, trials lasted 89.8 ± 17.0 min in duration. And for all lotion trials (TBC, TB, and PLAC), participants allowed their assigned creams to absorb a similar amount of time before beginning each trial (74.4 ± 11.0.2 min; p = 0.776).

3.2.2. Trial Average Power (Laps 2–4)

Results of a 4 × 3 repeated measures ANOVA produced significant main effects for differences by trial condition (p < 0.001, η²p = 0.929) with post hoc pairwise comparisons showing that FAM trials were significantly lower (p < 0.05) than TBC, TC, and PLAC trials. However, no differences (p > 0.05) were observed across laps based on trial condition (p > 0.05), thus there were no significant differences in average power for laps 2–4 (p = 0.316).

3.2.3. Trial Average HR and RPE (Laps 2–4)

ANOVA produced non-significant differences for both HR by trial condition (p = 0.782) and RPE (p = 0.499).

3.3. Hill Climb Segment Charactersitics and Outcomes

Hill Climb Segment Average Power (Laps 2–4)

Hill climb segments were ridden at 131.6 ± 21.1% of MMP60 ANOVA produced significant differences in average hill climb by trial condition (p < 0.001, η²p = 0.606), with post hoc pairwise comparisons showing that FAM trials were significantly lower (p < 0.001) than TBC, TC, and PLAC trials, while TBC trials were significantly higher (p = 0.038) than TC trials hill climb at 7.7 W. There were no significant differences between TBC and PLAC (p = 0.697) or TC and PLAC (p = 0.310).

3.4. Sprint Segment Charactersitics and Outcomes

Sprint Segment Average Power (Laps 2–4)

Sprint lap segment average power (laps 2–4) were ridden at 86.4 ± 7.3% of MMP30s. ANOVA results indicated significant differences in sprint power by trial condition (p < 0.001, η²p = 0.843), where post hoc comparisons revealed significantly lower sprint power for FAM trials compared to TBC trials (p < 0.001) and PLAC trials (p = 0.007), but not TC trials (p = 0.131). No significant differences were noted between TBC and TC trials (p = 0.259) or TC and PLAC trials (p = 0.725).

3.5. Time Trial Charactersitics and Outcomes

3.5.1. Time Trial Average Power (Lap 5)

The lap 5 TT was 15.7 ± 3.1 min in duration and ridden at 95.4 ± 8.7% of MMP20 and 87.5 ± 7.4% of MHR. Lap 5 TT power was not significantly different (p = 0.697) across trial conditions.

3.5.2. Time Trial Average HR and RPE (Lap 5)

Result of ANOVA indicated significant main effects for TT HR (p < 0.001, η²p = 0.994) but no significant individual effects (p > 0.05). RPE results also indicated significant main effects (p < 0.001, η²p = 0.989) where FAM trial RPE was significantly lower than TC trial RPE (p = 0.026) with no other significant differences. Overall trial RPE was not significantly different between trial conditions (p = 0.068); FAM = 6.8 ± 1.1, TBC = 7.0 ± 1.6, TC = 7.2 ± 1.3, PLAC = 6.6 ± 1.6.

3.6. Placebo Effects

For the purposes of this study, we considered accurate identification of supplement and PLAC as those individuals who correctly identified all three correctly; only two individuals achieved this, and one admitted he had used one of the products in the past. Overall, from the 44 reported total applications, 25 “believed” they received a real supplement with 20 of those being correct; however, in the results of Fisher’s exact test comparison, participants were equally likely to accurately identify a lotion correctly (p = 0.101), as well as identify any of the three lotions as a supplement (p = 0.922); see supplemental PLAC data for more results. As a secondary analysis, however, we wanted to examine whether trials where individuals believed they received an actual supplement performed better than those who believed they received a PLAC. Here, those who believed they received a real supplement (“Yes” = 24 vs. “No” = 19) performed similarly across all aspects of each trial with the largest, albeit non-significant difference in the Lap 5 TT; 233.7 ± 69.5 W vs. 221.5 ± 72.8 W (p = 0.579). While not significant, there was a mean advantage ranging from 2.4% in sprints to 5.5% in the TT, with wide variations (CV% = 20.7 vs. 23.9%) across participants including several negative responders; Supplement S4, Placebo related analyses, presents extensive graphic representations of PLAC and supplement ratings data.

4. Discussion

The purposes of this study were to investigate the ergogenic effects of both TBC and TC on a 50 km cycling trial that included three 300 m sprints and three 900 m hill climbs, as well as a final 9.1 km TT using the Zwift MR cycling platform. Participants applied the lotions based on instructions provided by the company and completed the supplement lotion trials in a randomly determined order Based on the available research, the authors hypothesized that neither TBC nor TC would elicit any measurable performance, HR, or RPE differences from individuals provided with similar a mentholated PLAC.

4.1. Trial Data

We believe our trials were able to test whether the claims made for these marketed products improved performance for high intensity repeated and sustained efforts, as well as overall endurance. Zwift’s MR platform combined with a smart bicycle trainer allowed participants to complete a 90–120 min trial that included short steep climbing efforts at ~132% MMP60 and long sprinting efforts at ~83% of MMP30s over nearly 30 km before concluding with a 10 km TT performed at over 95% MMP20, 90% MHR, and an RPE over 8. Indoor trials provided a controlled thermal environment where weight loss was negligible, indicating dehydration did not play a significant role in performance.

4.1.1. Laps 2–4 Repeated Efforts and Endurance

As noted, each trial consisted of three laps that included one 900 m steep climb ridden as fast as possible, followed by 4 km of steady riding before a 300 m sprint. The purpose was to examine whether either TBC or TC, buffers, outperformed a PLAC regarding higher average power for these six high-intensity intervals. Based on our results there was significant improvement in performance from FAM trials but largely negligible improvement in hill climb or sprint power. Specifically, average hill climb power for our TC group was about 8 W lower than the other condition groups, which could be accounted for daily variation [3] or small accuracy differenced in the bicycle trainer used [8]. Similarly, Wingate-type sprints, like we used, can vary as much as 10% [51], thus the 4.8% difference between our PLAC and TC group is reasonable. Additionally, there were no discernible differences average lap power, HR nor RPE.

4.1.2. Durability of Time Trial Performance

Cyclists who exhibit greater durability during events are more likely to succeed [15,17]; therefore, any ergogenic aid that can reduce fatigue and preserve MMP would improve one’s final TT durability. Based on the marketing claims of both TBC and TC, we should expect an improvement in 10 km TT power output following ~30 km of “moderate-hard” cycling (RPE ~6.4) with six hard intervals. However, our findings indicate that neither TBC nor TC offer any ergogenic benefit to performance with both power, HR, and RPE, which are remarkably similar across all TT’s, which were ridden at nearly MMP20.

Taken together, these data align with the majority of past research on both TBC [29,30,31,32] and TC [34]. However, close inspection of the other studies does not lend strong support for performance enhancement of repeated sprints, as one study showed only a trivial 1.9% improvement for TBC [28], while another using TC [35] showed significant increases in peak power for just three of twelve 6 s bike sprints in rugby players, but no differences in average 6 s power. Finally, an early study in elite soccer players using TC [33] showed significant improvements in 1000 m running TT performance but not repeated sprints. However, significant methodological issues call into question the validity of these data, including a lack of randomization, blinding, or placebo lotions, where TC trials were conducted last and produced improvements as great as 15% in one subject.

4.2. Performance Replication and Perceived Effort

An important aspect of this study was its relative difficulty, including several high-intensity repeats of 30 s to 3 min in duration, a final maximal TT, and an overall trial time of 90–120 min. Across all trials we saw no significant difference in session RPE which ranged from 6.6 for FAM, up to 7.1 for TC trials. Nonetheless, our participants demonstrated an ability to produce consistent performances after just a single familiarization trial, which also supports prior work using MR platforms like Zwift [48,52]. Thus, we are confident that these data demonstrate the quality of our study design and resulting conclusions.

4.3. Placebo Effects and Menthol

One of the challenges of supplement studies are potential PLAC (or nocebo) effects that have been shown to result in moderate effects which could result in 5% or greater performance changes [34,39,53,54]. In the present study, participants were asked whether they believed they received the supplement or PLAC, as well as providing qualitative and quantitative feedback on their opinions of the supplement. There were no overall differences in any performance measures between those who believed they received the actual supplement, though there appeared to be large variations across all subjects indicating that individual responses to PLAC effects could provide significant benefit to those who believe a supplement works. One interesting finding, however, was that supplement ratings and potential performance effects were independent of the ratings, where TBC was rated significantly lower than the other two lotions. In other words, individuals do not need to “like” or “prefer” a supplement for it to be effective. However, this may not apply to nocebo’s, as an earlier study on TC by the lead author [34] noted that the four negative responders all reported a significant dislike of menthol; it is noteworthy that no participants in the present study expressed negative comments regarding menthol.

An interesting unintended set of results from this study were the varying concentrations (dose) of menthol received from each lotion, ranging from 0.8% for TBC, 1% for P, and 1.25% for TC. In addition to lack of effect from supplement ingredient and PLAC effects, there also did not appear to be any effect from topical application of menthol across dosages either. This finding supports prior research showing no performance enhancement when apply topical menthol in a range of conditions [55,56,57]. One of the earlier TBC studies [28] suggested that the interaction between the menthol and TBC could explain their reported significant 1.9% improvement in repeated sprint performance. However, this seems unlikely based on the literature and the present study, in particular. Moreover, performance gains of less than 2% could be accounted for by random daily variation or a PLAC effect, noting that our smallest, and non-significant, PLAC effect was 2.4%.

4.4. Does Transdermal Delivery Make Sense for Buffers?

There is great appeal for transdermal application over either oral or hypodermic injection [58]. Specific to either bicarbonate or carnosine, ingestion can either be potential detrimental due to GI distress [25] or an inability to absorb the supplement in the gut [18,27], respectfully. Therefore, transdermal application would offer a convenient delivery route. As is typical for first and second-generation transdermal delivery systems, however, both TBC and TC are limited by the inherent challenges of skin absorption, as both topicals are hydrophilic and delivered in high doses, while carnosine is a peptide [58,59]. Based on the available ingredient lists, both products presumably attempt to enhance absorption using lipid and/or alcohol-based chemical enhancers. Nonetheless, evidence for absorption of either product is limited.

A recent paper by Gibson et al. [26] indicated a significant change in pH utilizing a TBC dosage 80 mL (~26.4 g of BC), which was only slightly more than most of the TBC performance research [28,29,31]. However, the small sample size, discrepancies between the paper and the raw data (provided by the authors), and the lack of an exercise challenge make conclusions impossible. Results from Kern et al. [32] showed a small, but significant change on blood acid-base markers but no performance improvements. What is intriguing, however, is that while Gibson et al. used a dosage four-times higher than manufacture recommendations, Kern et al. used a dosage of just 16.6 mL (5.5 g BC), which was also less than the 20 mL used in the present study. Therefore, based on the available evidence, we cannot say whether TBC alters blood buffering capacity.

Unlike TBC, absorption of TC is limited to a single study on equine published by Dieter et al. [27] indicating that application of the same TC used in the present study increased intramuscular carnosine concentrations by 46%. While this increase is on par with chronic oral BA ingestion [18], there is cause for skepticism. First, it is unclear what dosage of TC was used in the study. The authors note that the TC product used contained a 1.5% carnosine-complex concentration, which as used in the present study would be 0.15 g of carnosine in 10 mL of gel. Assuming this is an effective dose for a 70–80 kg human, then a dose six-fold higher should be needed for a 500 kg horse. As noted in the paper, however, the lead author Dieter has direct ties to Velocity Animal Sciences, which markets an identical TC product that advises an application of 40 to 60-mL should be used [60]. The authors state that the gel was applied to a 25-cm² area; this is implausible because just 10-mL of this gel easily coats ~3800-cm² (i.e., both lower extremities). Thus, based on the limited evidence, we cannot know whether TC is absorbed into the muscles.

4.5. Assimilation of Present Finding with Prior Transdermal Research

As noted above, the present study failed to show any indication of performance improvement from either TBC or TC when compared to PLAC. This finding was not surprising based on all the published research we could identify, including this study; these studies are summarized in

Table 3. Summary of current and prior TBC and TC supplement studies.

Author	Treatment	Sport	n	RCT	PLAC	Matched PLAC	Measures	Improvement	DB Score (30)
Harnish et al. 2025	TBC/TC	Cycling	15	Yes	Yes	No	RS, HI, TT, END	No improvement across all measures.	26
Gurton et. al. 2024 [29]	OBC/TBC	Soccer	10	Yes	Yes	No	RS	No change in acid-base balance and no improvement across all measures for TBC.	24
Gurton et. al. 2023 [28]	OBC/TBC	Team Field Sport	14	Yes	Yes	No	RS	No change in acid-base balance, similar < 2% improvement for RS compared to OBC.	24
McKay et al. 2020 [31]	OBC/TBC	Cycling	10	Yes	Yes	No	RS	No change in acid-base balance and no improvement across all measures for TBC.	23
Kern et al. 2019 * [32]	TBC	Cycling	20	Yes	Yes	No	S, TT	No change in acid-base balance and no improvement across all measures for TBC.	22
Seah et al. 2019 ^ [30]	TBC	Team Field Sport	10	Yes	Yes	?	RS	No improvement across all measures.	6
Brockelbank 2024 # [35]	TC	Rugby	12	Yes	Yes	Yes	RS	No improvement in average peak or mean sprint power.	24
Harnish and Miller 2023 [34]	TC	Cycling	15	Yes	Yes	No	RS	No improvement across all measures.	28
Sharpe and Macias 2016 [33]	TC	Soccer	11	No	No	No	RS, TT	Yes. Final TC trials showed small TT improvement.	12

* Unpublished manuscript; ^ Abstract only; ^# Unpublished manuscript plus dataset. DB: Downs and Black score; RCT: randomized controlled trial; RS: repeated sprint; S: sprint; HI: high-intensity intervals; END: endurance trials ≥ 90 min.

with primary findings and a rated for quality using the Downs and Black methodological quality checklist [61]. Six of these studies tested TBC and 4 TC, with studies comparing TBC to oral BC (OBC), and only the present study comparing TBC and TC with PLAC. Four of the TBC and one of the TC studies reported using a matching PLAC; however, based on our own experience with both products and unpublished data testing TBC and TC blinding, it is unlikely any lotion with BC would feel or apply like an identical lotion without BC. Five of the studies were industry funded, but only one [33] failed to use either a randomized trial or identifiable PLAC. Three studies were not peer-reviewed [30,32,35] and one had no detailed data published [30]. We were, however, able to obtain an unpublished manuscript for Kern et al. [32] and the complete dataset for Brockelbank [35], but we received no response from the other TC-related studies [27,33]. From these data, we were able to assess the quality of the reported findings and confirm their results. We note that despite the small (<2% improvement) seen in repeated sprint performance, the results were well within the range of a PLAC effect [39,53], or as noted by the Gurton et al. (2003) [28], possibly due to the menthol in the product, as there was no evidence of change in acid-base balance. Finally, based on the limited details provided about their study design, gaps in their reported results, and the rather remarkable individual responses of some of their subjects (>10%), we fundamentally question the conclusions by Sharpe et al. that the TC lotion employed enhances sport performance [33]. Taken together, we believe there is insufficient evidence that either TBC or TC provided any performance enhancement in either intermittent sprint or endurance sport conditions.

4.6. Novelty and Limitations

This is one of just a few studies to adopt a remote approach to conducting an RCT, as well as one of growing number of studies utilizing MR platforms like Zwift [10]. as well as one of growing number of studies utilizing MR platforms like Zwift [10]. Such virtual environments have been shown to produce reliable cycling performances [48], creating exciting new opportunities to conduct large scale cycling trials in semi-controlled environment on virtual riding routes. At the time of the study design, it was believed that such a platform would expand the potential subject pool. However, converting interested participants into enrolled and completed participants proved to be very challenging. While MR remote studies hold promise, a significant limitation of this and future remote studies are barriers to the collection of physiological data like VO₂, blood lactate, or markers of acid-base balance; resulting in the loss of insightful data that aids the interpretive process. Nonetheless, the primary purpose of the study was to test marketing claims against actual performance outcomes in a challenging trial. Finally, we acknowledge that our study lacked blinding of the research team and employed supplements that were easily distinguishable from one another. However, the impact of this limitation may have been of little consequence as most participants (13/15) failed to accurately identify all the lotions correctly, with more than half preferring or identifying the PLAC as a supplement as well.

4.7. Applications and Implications

The major applications and implications of this study are that neither TBC nor TC appear to have any discernable performance effect beyond a potential and highly variable PLAC effect. Thus, athletes and coaches should use these results to make informed choices on whether a product is worth the expense of purchase. The authors cannot discount the potential benefits of a PLAC effect in sports and thus make no specific judgment or recommendation for or against the use of any of these lotions, including the PLAC, to achieve that effect. Finally, as previously noted, our participants demonstrated that a single familiarization trial may be sufficient for even novice cyclists to perform replicable performance trials. This should encourage researchers to employ the use of MR platforms, like Zwift, for future research.

5. Conclusions

In conclusion, this study failed to show any evidence of performance enhancement for either TBC or TC lotions compared to PLAC in a group of trained cyclists completing a moderate duration endurance trial that included several repeated high-intensity efforts and a final time trial. Furthermore, when these results are taken together with prior research, we are left with the inconvenient truth that transdermal buffering lotions fail to perform as marketed. However, the study does support the feasibility of MR platforms like Zwift for the completion of unsupervised, remote cycling-performance experimentation over a variety of virtual terrain settings.