Discrepancy of Beta-Hydroxybutyrate Measurements between a Blood Meter and GC-MS Methods in Healthy Humans

Round 1

Reviewer 1 Report

In the present work the authors presented an interesting research with relevant results regarding the measurement of beta hydroxybutyrate.

The manuscript reads well with a clear presentation of the methodological approach and results obtained.

The only comment is related to the main topic covered by the authors in relation to the journal scope. Although monitoring ketosis can be considered a relevant topic fitting under the umbrella of muscle physiology studies, the authors may consider to include some additional sentence describing some example of how their results should be considered in light of studies performed on this field or similar (e.g., ketosis and muscle physiology) and how to consider their results for future studies in this field.

Please consider this just as possible suggestion, as the manuscript, is already good in its current form in my opinion.

Author Response

Comment 1: The only comment is related to the main topic covered by the authors in relation to the journal scope. Although monitoring ketosis can be considered a relevant topic fitting under the umbrella of muscle physiology studies, the authors may consider to include some additional sentence describing some example of how their results should be considered in light of studies performed on this field or similar (e.g., ketosis and muscle physiology) and how to consider their results for future studies in this field. Please consider this just as possible suggestion, as the manuscript, is already good in its current form in my opinion.

Response: Thank you for your suggestion to strengthen our paper as a better fit for Muscles.  We have added the following to the discussion:

Lines 237-254: “βHB can be viewed as short-chain water soluble fatty acids as providing 2 Acetyl-CoA molecules per βHB molecule. This makes βHB particularly advantageous for skeletal muscle during exercise when ATP requirements are high and there is a demand for quickly accessible energy sources. Since βHB is water soluble and its entry into mitochondria is not controlled by CPT-1 system, it becomes an important source of energy during exercise, thus sparing glycogen and potentially improving performance. During endurance exercise, βHB was shown to be re-leased from the liver to supply working muscles [11]. Ketone metabolism adaptations occur in the skeletal muscles after regular endurance exercise training; specifically, skeletal muscle increases its capacity to utilize βHB for fuel which leaves less βHB in the bloodstream post exercise [12,14]. Sedentary individuals have substantially higher levels of circulating βHB after strenuous pro-longed exercise than trained [12]. Trained individuals may find it difficult to maintain a βHB blood level associated with ketosis (0.5-3 mM) due to their enhanced capacity to utilize the βHB for fuel [13,22,23]. Ketone supplements are often used to increase the circulating supply of βHB to assist in maintenance of ketosis. Blood meters are used to determine if ketosis is achieved, however, our results demonstrate that βHB levels may be higher than the blood meter reports, regardless of supplementation. Further explorations should identify the relationship between results from the blood meter and GC-MS at various blood βHB levels and synchronize levels associated with ketosis for each measurement method.”

Reviewer 2 Report

The manuscript is clear and presented in a well-structured manner, and the references are most important for this research field. The research design is appropriate to test the hypothesis, but some concerns exist regarding the data analysis and sample size. Some clarifications and further data analysis should be developed.

 

 

Specific comments:

 The rationale of the study should be enhanced in the introduction so that we can understand the relevance of this study.

The authors stated, “Two subjects with at least one missing data point for the βHB analysis were removed from all analyses via listwise deletion. Outliers were operationally defined as data points with a standardized value >3.0 from the group mean.” Please, clarify the initial sample size and how many participants were removed. Please include the final sample size and characteristics.

The authors should provide more details on sample characteristics so that they can be reproducible and divided into the groups analyzed.

Limitations should be improved. More information on limitations should be added, such as sample size per group and characteristics.

More data analysis should be performed so that the results can be more robust. More details on differences and errors between measurements. It would be positive for readers to understand individual differences (for example, individual plots) and/or agreement (for example, Bland Altman plots). This is mandatory for further analysis of the paper.

 

 

Author Response

Comment 1: The rationale of the study should be enhanced in the introduction so that we can understand the relevance of this study.

Response: Thank you, the following text was added to the introduction to improve understanding of the rationale and aims:

Lines 76-84: “As most of the population, including clinicians and researchers, typically measures physiological levels of βHB with a blood meter, it is important to ensure the validity of the method as well as to clarify that the resulting quantity consists of the D-βHB isomer only, especially when assessing blood βHB levels after consuming a racemic ketone supplement which contains both D- and L-βHB. Therefore, the primary aim of this study was to investigate the efficacy of a blood meter to measure serum βHB in comparison with GC-MS following consumption of a racemic KS supplement or a placebo. The secondary aim was to investigate the efficacy of a blood meter to measure endogenous serum βHB in comparison with GC-MS in people in a fasted and resting state prior to any supplementation.”

 Comment 2: The authors stated, “Two subjects with at least one missing data point for the βHB analysis were removed from all analyses via listwise deletion. Outliers were operationally defined as data points with a standardized value >3.0 from the group mean.” Please, clarify the initial sample size and how many participants were removed. Please include the final sample size and characteristics.

Response: In section 4.2 we have addressed this issue by clarifying that there were 16 participants initially recruited for the study, but that 2 of those initial participants were removed from analysis due to data loss:

Lines 299-301: “Data loss for two of the participants resulted in their removal from the statistical analysis. Therefore, results for only 14 subjects are reported. Characteristics of the 14 participants with complete data are provided in Table 1.”

Comment 3: The authors should provide more details on sample characteristics so that they can be reproducible and divided into the groups analyzed.

Response: Near section 4.2 we have added a table (Table 1) to display the age, height, weight, and body mass index of the total sample, as well as the respective values for men and women:

	Men (n = 7)		Women (n = 7)		Total (N = 14)
	M	SD	M	SD	M	SD
Age (years)	22.57	(5.68)	20.43	(1.40)	21.5	(4.17)
Height (cm)	186.50	(11.58)	165.05	(10.93)	175.77	(15.52)
Weight (kg)	83.67	(13.71)	72.90	(22.44)	78.28	(18.72)
BMI (kg/m2)	24.43	(5.93)	26.53	(6.78)	25.48	(6.21)

Comment 4: Limitations should be improved. More information on limitations should be added, such as sample size per group and characteristics.

Response: Several limitations of the study were noted in the final paragraph of Discussion (section 3), but we agree that there are more limitations that are worthy of discussion when interpreting this study. The following text was added to the limitations paragraph to discuss the limitation noted above and explain that future studies should consider using a larger and more diverse sample to increase generalizability of the results:

Lines 269-272: “A final limitation of this study to consider is the sample size of 14 participants. Achieved statistical power was not a concern (Power >.81 for all ANOVA results), but the relatively low sample size limited the potential variability of the participants. A larger and more diverse sample for future studies may yield results that are more applicable to the general population.”

 Comment 5: More data analysis should be performed so that the results can be more robust. More details on differences and errors between measurements. It would be positive for readers to understand individual differences (for example, individual plots) and/or agreement (for example, Bland Altman plots). This is mandatory for further analysis of the paper.

Response: Thank you for the constructive recommendation. We have updated Figure 1 to include individual data points and lines indicating changes in measures between time points for each individual participant, as well as group differences noted in the text.

We have also included Bland-Altman plots (Figure 2) for each supplement condition and time point (4 plots total). The graphs in this figure indicate the bias (average error) between devices and the agreement between devices at varying measurement levels. In addition to the Figure and corresponding legend, the following addition to section 4.5 has been made:

Lines 361-367: “Bland-Altman plots were constructed for blood βHB measurements taken using blood meter and GC-MS to assess agreement between devices. At each time point (PRE and POST) and in each condition (placebo and KS) the readings for each device were used to calculate the bias (average difference between measurement devices) and upper and lower limits of agreement (bias ±2SD of bias) between devices for measuring βHB across the range of observed values. The bias and limits of agreement were then used to construct the Bland-Altman plots to visually assess agreement between devices at each time point in each condition.”

Additionally, reference to, and interpretation of, Figure 2 has been added in the Results (section 2) and the Discussion (section 3):

Results, Lines 114-118: “There was a substantial level of bias between the blood meter and GC-MS methods of measurement as displayed in Figure 2. Bias ranged from 0.52 – 2.64 mmol/L depending on the condition. The level of variability in measurement between methods (upper and lower limits of agreement) was large relative to the level of measurement of βHB. The trend lines for each plot show an increasing level of disagreement between devices (i.e., more error) as the level of βHB increases.”

Discussion, Lines 202-204: “The large differences between devices and increasing error with higher βHB values displayed in Figure 2 emphasize these measurement discrepancies between devices.”

Next, we have added a graph (Figure 3) to display the difference in gain scores for each device following ketone salt ingestion. This analysis was described in the text, but the differences between devices with individual data points are now included in Figure 3. This figure (and accompanying text) also includes a comparison between the blood meter and 50% of the GC-MS gain following ketone salt supplementation to more fully interpret the differences in βHB that were found.

The following text was added to the Results (section 2) and Discussion (section 3):

Results, Lines 133-144: “After the formal analysis of the results, a post hoc gain-score analysis (comparing the changes in βHB value between blood meter and GC-MS from PRE to POST) was deemed appropriate to further investigate the nature of the differences in exogenous βHB measurement between devices. A significantly greater increase in βHB was detected when GC-MS (2.36 ± 2.25 mmol/L) was used compared to when blood meter (0.24 ± 0.31 mmol/L) was used (p = 0.003, g = 0.92). An additional gain score comparison was made between the blood meter and 50% of the value attained from the GC-MS (50% GC-MS) to account for the fact that 50% of the KS supplement was of the L-isomer, which cannot be measured by the blood meter. Thus, computing the gain score adjusts for initial endogenous βHB and halving the GC-MS gain value adjusts for the L-isomer that is included in GC-MS but not blood meter readings. βHB was greater in 50% GC-MS (1.18 ± 1.12 mmol/L) compared to the blood meter (p = .006, g = .831). These gain scores are represented graphically in Figure 3.”

Discussion, Lines 192-197: “A secondary gain score analysis determined that when the gain in total βHB detected by GC-MS was halved (to account for the fact that half of this βHB increase was undetectable by blood meter) the resulting increase in exogenous βHB was still significantly and substantially greater than that detected by blood meter. The GC-MS sample analysis method used prevented the precise measurement of D-βHB, yet this estimation provides some evidence that the difference in βHB assessment between devices was not solely due to the inclusion of L-βHB.”

We believe that these figures and their explanation contribute to the interpretation of the data and strengthen the conclusion of the paper.

In addition to these new and updated graphs, we have added effect size estimates to all significant differences reported in the Results (section 2) to report the results of the study more completely.

Reviewer 3 Report

The current study compared beta-hydroxybutyrate concentrations in serum from human subjects consuming a placebo or ketone salt supplement using a commercially available meter or gas chromatography/mass spectrometry (GC/MS).  The findings report that serum ketone body concentrations increased 30 minutes after consuming the ketone salt supplement.  Furthermore, values obtained with GC/MS are significantly higher than the blood meter. 

 

Comments, Concerns, and Suggestions:

 

The purpose and relevance of the study is not clear.  The authors are comparing the blood meter, which is known to measure D-BOHB (the biological relevant enantiomer), with GC/MS which measures both the D- and L- enantiomers.  Since the method described in the paper is unable to discriminate between the D- and L-, the findings are quite obvious and expected.  Ketone salts are known to provide both enantiomers, which is a criticism regarding their use to increase ketosis.  The authors state that they are not able to measure the concentration of D- and L- and do not know the relative concentration of each enantiomer in the provided supplement, so the suggested claim that blood ketone meters are accurate are unsupported with direct evidence. 

 

Why do the authors imply that the GC/MS method is the “gold standard”?  The paper referenced in line 71 (Ref #21) achieved similar values to commonly used clinical tests.  Furthermore, the authors suggest the use of their method for forensic analysis.  The sample preparation in the referenced paper are different than the what the authors used.  Have the authors performed any precision analysis to help validate their method?

 

 

The study would be improved significantly if the concentration of the D- vs L- enantiomers could be determined.  There appears to be LC-MS methods that can do this.  Further comparison between the method and other standard methods (i.e., what is used clinically?) would be helpful.  The inclusion of comparison using a ketone body ester would be important as well.

 

Author Response

Comment 1: The purpose and relevance of the study is not clear.  The authors are comparing the blood meter, which is known to measure D-BOHB (the biological relevant enantiomer), with GC/MS which measures both the D- and L- enantiomers.  Since the method described in the paper is unable to discriminate between the D- and L-, the findings are quite obvious and expected.  Ketone salts are known to provide both enantiomers, which is a criticism regarding their use to increase ketosis.  The authors state that they are not able to measure the concentration of D- and L- and do not know the relative concentration of each enantiomer in the provided supplement, so the suggested claim that blood ketone meters are accurate are unsupported with direct evidence. 

Response: Thank you for your comments. The purpose of the study was to determine the extent to which the blood meter underestimates ketosis induced by the ketone salt ingestion as compared with the “gold standard”, i.e., GC/MS. We clarified this in the text. To the best of our knowledge, such comparison has not been done and we felt that it is important to address this issue. The KS mix we used has a racemic mixture of D and L enantiomers, thus, every subject received about 67 mmoles of each isomer or about 1.7 mmole/kg in total or 0.8 mmole/kg of D-BHB.

The following text was added to the introduction to improve understanding of the rationale and aims:

Lines 76-84: “As most of the population, including clinicians and researchers, typically measures physiological levels of βHB with a blood meter, it is important to ensure the validity of the method as well as to clarify that the resulting quantity consists of the D-βHB isomer only, especially when assessing blood βHB levels after consuming a racemic ketone supplement which contains both D- and L-βHB. Therefore, the primary aim of this study was to investigate the efficacy of a blood meter to measure serum βHB in comparison with GC-MS following consumption of a racemic KS supplement or a placebo. The secondary aim was to investigate the efficacy of a blood meter to measure endogenous serum βHB in comparison with GC-MS in people in a fasted and resting state prior to any supplementation.”

The following text was added to the methods section regarding the βHB isomers in the KS supplement:

Lines 292-293: “KS consisted of 7 grams of racemic sodium D, L-βHB (50% D- βHB and 50% L- βHB) and flavoring mixture and the placebo consisted of maltodextrin, sodium, and flavoring.”

In addition, Figures 2 and 3 have been added and Figure 1 updated to better clarify the purpose and findings of this study.  Regarding Figure 3, the 50% Gain score was performed to help account for the inclusion of L-BHB in GC-MS.  The following text to describe this was added to the Results (Section 2) and Discussion (section 3):

Lines 133-144: “After the formal analysis of the results, a post hoc gain-score analysis (comparing the changes in βHB value between blood meter and GC-MS from PRE to POST) was deemed appropriate to further investigate the nature of the differences in exogenous βHB measurement between devices. A significantly greater increase in βHB was detected when GC-MS (2.36 ± 2.25 mmol/L) was used compared to when blood meter (0.24 ± 0.31 mmol/L) was used (p = 0.003, g = 0.92). An additional gain score comparison was made between the blood meter and 50% of the value attained from the GC-MS (50% GC-MS) to account for the fact that 50% of the KS supplement was of the L-isomer, which cannot be measured by the blood meter. Thus, computing the gain score adjusts for initial endogenous βHB and halving the GC-MS gain value adjusts for the L-isomer that is included in GC-MS but not blood meter readings. βHB was greater in 50% GC-MS (1.18 ± 1.12 mmol/L) compared to the blood meter (p = .006, g = .831). These gain scores are represented graphically in Figure 3.”

Discussion, Lines 192-197: “A secondary gain score analysis determined that when the gain in total βHB detected by GC-MS was halved (to account for the fact that half of this βHB increase was undetectable by blood meter) the resulting increase in exogenous βHB was still significantly and substantially greater than that detected by blood meter. The GC-MS sample analysis method used prevented the precise measurement of D-βHB, yet this estimation provides some evidence that the difference in βHB assessment between devices was not solely due to the inclusion of L-βHB.”

Comment 2: Why do the authors imply that the GC/MS method is the “gold standard”?  The paper referenced in line 71 (Ref #21) achieved similar values to commonly used clinical tests.  Furthermore, the authors suggest the use of their method for forensic analysis.  The sample preparation in the referenced paper are different than the what the authors used.  Have the authors performed any precision analysis to help validate their method?

Response: As evident from our results, blood meter significantly underestimated ketone body concentrations after ketone salt ingestion and even at baseline, consistent with other works (e.g., Stubbs et al, Ref 17). To evaluate the extent of underestimation, we used GC/MS because it is able to discriminate between acetoacetate and beta-hydroxybutyrate in complex matrix such as blood and precisely measure all the extracted molecules. GC/MS remains “golden standard” because of its high accuracy, molecule specificity, and repeatability. The goal of this manuscript is not to present a novel methodology. We compared two existing well-established methodologies. The reference used for the GC/MS method used in our paper was for headspace method which is incorrect and has been updated. That reference mentions forensic analysis usage. We did not suggest this in our manuscript.

Comment 3: The study would be improved significantly if the concentration of the D- vs L- enantiomers could be determined.  There appears to be LC-MS methods that can do this.  Further comparison between the method and other standard methods (i.e., what is used clinically?) would be helpful.  The inclusion of comparison using a ketone body ester would be important as well.

Response: We only focused on D-BHB since it is detected by both methods and is a biologically relevant isomer. In addition, L-BHB is eliminated slowly and undergoes different metabolic route altogether so separate investigation and analyses would be required. This is outside of the scope of this paper; however, we will consider this in the future since ketone salts commonly have both isomers.

Also, it would be of great interest to repeat this study using a ketone ester supplement.  This has been added as a possible future exploration in the discussion.

Lines 267-268: In addition, future explorations should compare blood meter and GC-MS measures of circulating βHB after consuming a ketone ester supplement.

Round 2

Reviewer 2 Report

I would like to congratulate the authors on their revised work. The authors have addressed all my questions and concerns. 

Author Response

Dear Reviewer,

Thank you for your feedback. 

Reviewer 3 Report

Although the authors have made minor improvements, the revisions have not been sufficient to demonstrate the significance and importance of the study.  The purpose of the study is compare a commercially available device to that of a GC/MS approach.  With this, the authors imply that the commercially available device is incorrect.  Why was the device (i.e., Abbott) selected?  Have issues or concerns been reported?  Since the two methods do not detect and/or measure the same thing (D vs. D+L) and the GC/MS is unable to differentiate the concentrations of D vs L, the relevance of the study remains unclear to me.   This is a major limitation of the study, which the authors identify in the manuscript. 

Overall, the significance and importance of the study remains low.  What new information does this provide to clinicians, researchers, and the general public?  The authors seem to recommend that GC/MS should be used routinely to measure ketone body concentrations.  What methods are currently used clinically?  Do the authors imply that this methods are incorrect and that GC/MS is the method that should be used?  Is this practical?  There is no data provided that demonstrates the the GC/MS method used in the study is valid or reliable.  

Author Response

Dear Reviewer,

Thank you for your feedback. Please find our responses to your comments below.

Comment 1: The purpose of the study is compare a commercially available device to that of a GC/MS approach.  With this, the authors imply that the commercially available device is incorrect. 

Response: There are two aims of the paper stated in the last paragraph of the introduction section, which were:

“to investigate the efficacy of a blood meter to measure serum βHB in comparison with GC-MS following consumption of a racemic KS supplement or a placebo. The secondary aim was to investigate the efficacy of a blood meter to measure endogenous serum βHB in comparison with GC-MS in people in a fasted and resting state prior to any supplementation.”

The primary purpose of the study does not imply that the commercially available device is ineffective or inaccurate in general; rather, this purpose refers to a specific scenario in which racemic ketone salts are consumed prior to measurement. This condition has not been evaluated and validated in the literature to date, thus making our study the first of its kind. Therefore, we are suggesting that the accuracy and effectiveness of the blood meter under these circumstances is unknown, not that it is assumed to be inaccurate.

The secondary purpose of this study was to investigate the effectiveness of the meter under resting conditions in a fasted state. This is done in part to establish a baseline for the discrepancy in BHB measurement between the two devices, so that any differences in measurement may be interpreted more accurately. It is also known that blood meters are not 100% accurate. Rather these devices, including the Abbott Precision Xtra, are commonly used as a point-of-care ketone body assessment tool in clinical situations for diagnosing ketoacidosis (extremely high ketone levels). They are deemed appropriate for this purpose, but when used outside of clinical settings (for tracking athletic performance, ketogenic diet interventions, etc.), a more accurate and precise reading may be necessary. The use of the blood meter we used in this study may not be accurate, and therefore effective, in these other situations. 

The following was added to the Introduction:

Lines 80-84: However, no studies to date have focused on investigating the validity of commonly used blood meters to measure BHB in healthy humans. Additionally, no studies have addressed the potential discrepancy in BHB measurement following consumption of racemic ketone salts, which are increasingly used by athletes and the general public to induce a state of nutritional ketosis for a variety of purposes.

Comment 2: Why was the device (i.e., Abbott) selected?  Have issues or concerns been reported? 

Response: To address this, the following was added to the introduction:

Lines 75-79: Blood meters, such as the Abbott Precision Xtra, are commonly used as a point-of-care ketone body assessment tool in clinical situations for diagnosing ketoacidosis. They are deemed appropriate for this purpose because, although not as accurate as reference methods, they provide nearly instantaneous results at a lower cost which allows clinicians to make critical healthcare decisions based on more frequent ketone readings [24].

In this sense, the Abbott Precision Xtra device that we used is an excellent tool for its intended purpose and feasibility. However, no issues or concerns have been reported regarding the use of the meter to measure racemic ketone salts because no studies to date have investigated this topic.

Comment 3: Since the two methods do not detect and/or measure the same thing (D vs. D+L) and the GC/MS is unable to differentiate the concentrations of D vs L, the relevance of the study remains unclear to me.   This is a major limitation of the study, which the authors identify in the manuscript.

Response: It is true that the inability of the blood meter to measure half of the ingested BHB, in the experimental supplement condition at the POST time point, was a limitation of the study that we identified. However, very reasonable efforts were made to accurately interpret the findings of the study despite this limitation by adjusting the GC-MS values at this one time point based on known supplement information. We included a new graph in Figure 3 which shows the change in BHB measured by each device, in addition to 50% of the change in the BHB detected by the GC/MS device, with the corresponding text in the Results section:

“An additional gain score comparison was made between the blood meter and 50% of the value attained from the GC-MS (50% GC-MS) to account for the fact that 50% of the KS supplement was of the L-isomer, which cannot be measured by the blood meter. Thus, computing the gain score adjusts for initial endogenous βHB and halving the GC-MS gain value adjusts for the L-isomer that is included in GC-MS but not blood meter readings. βHB was greater in 50% GC-MS (1.18 ± 1.12 mmol/L) compared to the blood meter (p = .006, g = .831). These gain scores are represented graphically in Figure 3.”

and the Discussion section:

“A secondary gain score analysis determined that when the gain in total βHB detected by GC-MS was halved (to account for the fact that half of this βHB increase was undetectable by blood meter) the resulting increase in exogenous βHB was still significantly and substantially greater than that detected by blood meter. The GC-MS sample analysis method used prevented the precise measurement of D-βHB, yet this estimation provides some evidence that the difference in βHB assessment between devices was not solely due to the inclusion of L-βHB.”

The extent to which the findings and conclusions of the study are limited by this measurement issue is not so great that they are useless. This limitation is seemingly negligible in the context of the purpose of the study which was to investigate the efficacy of a blood meter to measure serum βHB, not to determine the specific accuracy or agreement of the blood meter with GC/MS.

The results have meaning and are applicable to various populations, as discussed in the manuscript and in comment responses below.

Regarding aim 2, there was no L- βHB in circulation in the control condition or in the two baseline conditions, which allows the difference in βHB measurement ability not to be a concern as both devices measured 0 mM of L- βHB in these non-KS conditions. We know this because L- βHB is an exogenous ketone isoform that does not exist in the body unless it is consumed. This finding was discussed in the manuscript in paragraph 6 of the Discussion section:

“Differences between measurement devices were present even when no exogenous ketones were consumed; that is, even when the issue of L-βHB detection was not a concern. The differences between the blood meter and GC-MS in both PRE-conditions suggests that there is a measurement discrepancy between devices even when only D-βHB is circulating and in the absence of any ingested substance, extreme fast, or clinical condition. The large differences between devices and increasing error with higher βHB values displayed in Figure 2 emphasize these measurement discrepancies between devices.”

Comment 4: Overall, the significance and importance of the study remains low.  What new information does this provide to clinicians, researchers, and the general public?

Response: The new information produced by this study and the importance of it was discussed in the Discussion.  Please see the excerpts below:

Paragraph 5: “This is relevant to people who routinely test their ketone levels in the morning (e.g., someone in the initial stages of a ketogenic diet) to determine when they have become keto-adapted. A commonly used cutoff point for marking the onset of nutritional ketosis is a βHB value >0.5mM [22]. After a 10-hour fast, the average βHB value for subjects in this study was above this cutoff as measured by GC-MS, but below the cutoff when measured with a blood meter. A closer examination of the data showed that in the PRE-conditions, the result was a false negative for ketosis (a blood meter reading below 0.5 mmol/L with a GC-MS reading above 0.5 mmol/L) in 23 of 28 instances. The slightly lower sensitivity of the blood meter to detect endogenous βHB may lead to more false negative results for those interested in detecting nutritional ketosis and should be taken under consideration when using blood meters for this purpose.”

Paragraph 6: “People who test their βHB throughout the day to maintain a state of ketosis would benefit from recognizing this difference between devices. For example, βHB levels lower than expected during the day may prompt one to consume a greater volume of KS than is necessary or to restrict carbohydrate intake to even more stringent levels than required.”

Paragraph 7: “At the very least, consumers of KS should be aware of the proportion of L-βHB in the product that they use and understand that this isoform yields different benefits than D-βHB. Manufacturers and consumers will be better armed to correctly interpret βHB readings from blood meters with such information.”

Paragraph 8: “Ketone supplements are often used to increase the circulating supply of βHB to assist in maintenance of ketosis. Blood meters are used to determine if ketosis is achieved, however, our results demonstrate that βHB levels may be higher than the blood meter reports, regardless of supplementation.” 

To summarize these points, the findings we present are relevant to athletes and other people in the general public who are interested in achieving and/or maintaining a state of nutritional ketosis and who rely on blood meters to collect information about this and make decisions based on such information. The findings are particularly relevant to people who consume racemic ketone salt supplements to achieve nutritional ketosis, because a proportion of the advertised supplement (βHB) will not be reflected in blood meter readings, a fact that is known to researchers in this specific field but which is not profoundly obvious to people in the general public. Expectations about βHB assessments obtained from blood meters can be tempered based on the findings we present.

Comment 5: The authors seem to recommend that GC/MS should be used routinely to measure ketone body concentrations. 

Response: We apologize for the confusion on this point. We do not recommend that GC/MS should be used routinely to measure ketone body concentration. We also state in the conclusion that:

“The use of blood meters to measure βHB in the absence of KS supplementation should also be interpreted with some caution, as they appear to be less sensitive than GC-MS which is the most valid and precise method for βHB assessment. It may be more difficult to diagnose nutritional ketosis with a blood meter for this reason.”

We do not think that this point requires further clarification since it is stated very explicitly that we recommend that blood meter users be aware of the inaccuracies associated with blood meters to detect BHB, not that they are completely useless.

Comment 6: What methods are currently used clinically? 

Response: While GC/MS systems are expensive and challenging to operate, there are a number of academic and private centers which offer GC/MS measurements which can be successfully employed in a setting of a research study, similar to this study. Blood meters, such as the Abbott Precision Xtra used in this study, are used in clinical settings as point-of-care meters in which instantaneous results are needed at low cost and acceptable accuracy is lower.

Comment 7: Do the authors imply that this methods are incorrect and that GC/MS is the method that should be used?

Response: No. The scenario in which ketones are measured determines the most appropriate testing device. GC/MS should be used in some scenarios when time and expense permit, and blood meters should be used in others when some accuracy can be sacrificed for a substantially lower time and money burden. 

Comment 8: Is this practical? 

Response: It is practical in some situations and not practical in other situations, as mentioned in responses to earlier comments.

Comment 9: There is no data provided that demonstrates the GC/MS method used in the study is valid or reliable.  

Response: βHB GC/MS determinations have been extensively studied and published. All of the aspects of βHB measuring such as extraction, derivatization, fragmentation patterns, ionization efficiency, molecular mass to be monitored have been published in many works prior to this publication. Laboratory of Dr. Bederman, senior author of this manuscript, has extensive experience with isolation and detection of many metabolites of interest including βHB. He has a proven publication track record in metabolic research and metabolomics. As ketone bodies gained significant popularity in recent decades thanks to Atkins diet as one example, ketone body measurements are routinely done using GC/MS making this method both valid and reliable.