We thank Huhe and colleagues for their comments on our review titled “Experimental basis for generating nonhuman primate models of frontotemporal dementia and Alzheimer’s disease” [1]. Their main critiques were twofold: (1) That adult marmoset brains express both 3R and 4R tau isoforms [2], contrary to our interpretation based on Sharma et al. [3] which suggested 3R tau is negligible; (2) That although we highlighted the superior homology of AD GWAS risk genes between marmosets and humans (relative to mice), we did not cite supporting references.
Regarding the first point, since it is inherently difficult to prove a complete absence of a biological molecule, a quantitative discussion is essential. Sharma et al. [3] reported that the relative quantity of 3R tau in adult marmoset brains was below detection thresholds using cDNA library-based PCR and western blotting—the former being generally more sensitive. We thus considered 3R tau to be a minor isoform.
In contrast, Huhe et al. used RT-PCR and LC-MS/MS following trypsin or LysArg enzymatic digestion, along with western blotting and immunohistochemistry, to demonstrate 3R tau expression [2]. Importantly, they confirmed the linearity of their mass spectrometric quantification. However, the LysArg-based method appeared approximately twice as sensitive in estimating the 3R/4R ratio compared to the trypsin-based method (Figure 2G), potentially leading to over- or underestimation depending on which is more accurate.
It is unfortunate that the authors did not statistically analyze the 4R/3R ratio difference between humans and marmosets (Figure 2F), although their data clearly indicate a higher 4R/3R ratio in marmosets. This trend is corroborated by their Western blot results (Figure 4). They also demonstrated a small but significant amount of 3R tau in adult mouse brains, long thought not to express this isoform.
The observed difference in 4R/3R ratios between humans and marmosets resembles that caused by the Int10+3 mutation in the human MAPT gene [4], which aligns with species-specific differences in intronic splicing regulatory elements [3]. Hence, this issue is fundamentally one of quantification, and we believe both Huhe et al. and our group are correct in concluding that marmosets are superior to mice as models for frontotemporal dementia and Alzheimer’s disease. It would be informative to quantify the 4R/3R ratio in other nonhuman primates, such as macaques, as marmosets and macaques represent New World and Old World primates, respectively.
As for the second issue, the GWAS data we referenced originated from Bellenguez et al. [5]. We used UniProt (the Universal Protein Knowledgebase, 2025 edition) [6] to calculate interspecies homology. We apologize for not including these references in our original review.
Once again, we thank Huhe and colleagues for drawing attention to our article. We believe that developing marmoset models of neurodegenerative diseases will advance our understanding of disease mechanisms and serve as a vital bridge in translational research—facilitating near-clinical testing of novel therapeutics.
Takahiro Morito¹, Naoto Watamura¹˒², Hiroki Sasaguri¹, Taisuke Tomita³, Makoto Higuchi⁴, Hideyuki Okano¹˒⁵, Erika Sasaki⁶, and Takaomi C. Saido¹*
¹RIKEN Center for Brain Science, Hirosawa, Wako, Japan
2UK Dementia Research Institute, University College London, London, United Kingdom
3Graduate School of Pharmaceutical Science, University of Tokyo, Hongo, Tokyo, Japan
4Institute for Quantum Medical Science, National Institutes for Quantum Science and Technology, Chiba, Japan
5Department of Physiology, School of Medicine, Keio University, Tokyo, Japan
6Central Institute for Experimental Animals, Kawasaki, Kanagawa, Japan
[3] Sharma G, Huo A, Kimura T, et al. Tau isoform expression and phosphorylation in marmoset brains. J Biol Chem 2019; 294: 11433-11444.
[4] Watamura N, Foiani MS, Bez S, et al. In vivo hyperphosphorylation of tau is associated with synaptic loss and behavioral abnormalities in the absence of tau seeds. Nat Neurosci 2025; 28: 293-307.
[5] Bellenguez C, Küçükali F, Jansen IE, et al. New insights into the genetic etiology of Alzheimer's disease and related dementias. Nat Genet 2022; 54: 412-436.
[6] UniProt Consortium. UniProt: the Universal Protein Knowledgebase in 2025. Nucleic Acids Res 2025; 53: D609-D617.
Thank you for carefully reviewing our manuscript titled "Rapid Cognitive Assessment: Accuracy and Discriminant Validity of Mini-Cog and Process-Based Clock Drawing Test" published in JAD and for pointing out the miscalculations in the negative likelihood ratio (NLR) values. We are grateful for your expertise and the time you spent examining our work.
Upon reanalysis of our original data, we confirmed that the NLR values in Table 3 were indeed calculated incorrectly, with the numerator and denominator reversed. We deeply regret this oversight, which may have mislead readers. Enclosed, please find the corrected Table 3, with the revised NLR values highlighted in red. The corrected NLR for the Clock Drawing Test (CDT) in diagnosing MCI is 0.21, and for the Mini-Cog3, it is 0.24. The small difference of 0.03 indicates that both tests have similar performance in ruling out MCI. However, the Mini-Cog3 shows a significantly higher area under the curve (AUC) of 0.82 compared to the CDT's AUC of 0.77, demonstrating its superior diagnostic accuracy.
Regarding your suggestion that “in the sequence of testing, the CDT should be administered first, and if the CDT is normal, further testing with the Mini-Cog for mild cognitive impairment may not be necessary,” we would like to clarify. While the CDT can indeed rule out MCI, if the CDT is abnormal, the Mini-Cog3 should not be administered subsequently, as it already includes the CDT. The Mini-Cog3 assessment begins with immediate recall of three words, followed by the CDT, and concludes with delayed recall of the same words. This comprehensive approach, which evaluates both memory recall and executive functions, is completed within a brief 2-3 minutes, making it an efficient tool for clinical settings. Consequently, we advocate for the direct use of the Mini-Cog3 in community screenings for MCI and early-stage Alzheimer’s disease to enhance detection rates without incurring significant time costs.
Given the minimal difference in NLR values and the superior diagnostic accuracy and comprehensive assessment provided by the Mini-Cog3, we maintain our recommendation that the Mini-Cog3 is the preferred tool for the rapid and accurate detection of cognitive dysfunction in patients with MCI and Alzheimer's disease.
Once again, we thank you for your insightful feedback.
Sincerely, Qian Yang
Corresponding author: Defa Zhu, Department of Geriatric Endocrinology, The First Affiliated Hospital of Anhui Medical University, Hefei, Anhui, China. Email: zdfa0168@sina.com
Qihao Guo, Department of Gerontology, Shanghai Jiao Tong University Affiliated Sixth People’s Hospital, Shanghai, China. Email:
Thank you sincerely for taking the time to read and acknowledge our research. We are also grateful for your insightful guidance regarding the shortcomings of our study. We have thoroughly reviewed your letter and have reflected deeply on the issues you highlighted in the text. The ADNI database serves as a global specialized cohort for clinical research on Alzheimer's disease, offering a rich array of clinical and biomarker data. The project aimed to focus on the AD pathology and AD clinical performance, which could introduce selection bias and limit the generalizability to other clinical populations. Therefore, the next step is to validate our findings through studies that include more diverse populations. We whole heartedly support the author's proposal for standardizing ATN cutoff values. However, achieving this goal will necessitate collaborative efforts and extensive validation across multiple studies. We eagerly anticipate a future where standardized cutoff values for AD-related biomarkers are more widely available. Due to the absence of a frailty assessment program within the ADNI cohort and the lack of a standardized method for evaluating frailty, we have faced significant challenges in our assessment efforts. To address this, we have turned to the mFI-11 scale as applied by Soon et al. [1] in the ADNI cohort. We are optimistic about achieving a precise evaluation of frailty in this cohort in the future and plan to explore the replication of our findings using various assessment scales. We sincerely appreciate your valuable feedback on our research. Your insights will guide us in enhancing our study in future. Thank you once again.
Bao-Lin Han1, Hui-Fu Wang1,2,*, Lan Tan1,2,*
1Department of Neurology, Qingdao Municipal Hospital, Qingdao University, Qingdao, China
2Department of Neurology, Qingdao Hospital, University of Health and Rehabilitation Sciences (Qingdao Municipal Hospital), Qingdao, China
References [1] Soon SXY, Kumar AA, Tan AJL, Lo YT, Lock C, Kumar S, Kwok J, Keong NC (2021) The impact of multimorbidity burden, frailty risk scoring, and 3-directional morphological indices vs. testing for CSF responsiveness in normal pressure hydrocephalus. Front Neurosci 15, 751145.
The investigators did an excellent job studying a 40-year cohort to provide definitive population evidence that oral health is a risk factor for dementia.
We thank Huhe and colleagues for their comments on our review titled “Experimental basis for generating nonhuman primate models of frontotemporal dementia and Alzheimer’s disease” [1]. Their main critiques were twofold: (1) That adult marmoset brains express both 3R and 4R tau isoforms [2], contrary to our interpretation based on Sharma et al. [3] which suggested 3R tau is negligible; (2) That although we highlighted the superior homology of AD GWAS risk genes between marmosets and humans (relative to mice), we did not cite supporting references.
Regarding the first point, since it is inherently difficult to prove a complete absence of a biological molecule, a quantitative discussion is essential. Sharma et al. [3] reported that the relative quantity of 3R tau in adult marmoset brains was below detection thresholds using cDNA library-based PCR and western blotting—the former being generally more sensitive. We thus considered 3R tau to be a minor isoform.
In contrast, Huhe et al. used RT-PCR and LC-MS/MS following trypsin or LysArg enzymatic digestion, along with western blotting and immunohistochemistry, to demonstrate 3R tau expression [2]. Importantly, they confirmed the linearity of their mass spectrometric quantification. However, the LysArg-based method appeared approximately twice as sensitive in estimating the 3R/4R ratio compared to the trypsin-based method (Figure 2G), potentially leading to over- or underestimation depending on which is more accurate.
It is unfortunate that the authors did not statistically analyze the 4R/3R ratio difference between humans and marmosets (Figure 2F), although their data clearly indicate a higher 4R/3R ratio in marmosets. This trend is corroborated by their Western blot results (Figure 4). They also demonstrated a small but significant amount of 3R tau in adult mouse brains, long thought not to express this isoform.
The observed difference in 4R/3R ratios between humans and marmosets resembles that caused by the Int10+3 mutation in the human MAPT gene [4], which aligns with species-specific differences in intronic splicing regulatory elements [3]. Hence, this issue is fundamentally one of quantification, and we believe both Huhe et al. and our group are correct in concluding that marmosets are superior to mice as models for frontotemporal dementia and Alzheimer’s disease. It would be informative to quantify the 4R/3R ratio in other nonhuman primates, such as macaques, as marmosets and macaques represent New World and Old World primates, respectively.
As for the second issue, the GWAS data we referenced originated from Bellenguez et al. [5]. We used UniProt (the Universal Protein Knowledgebase, 2025 edition) [6] to calculate interspecies homology. We apologize for not including these references in our original review.
Once again, we thank Huhe and colleagues for drawing attention to our article. We believe that developing marmoset models of neurodegenerative diseases will advance our understanding of disease mechanisms and serve as a vital bridge in translational research—facilitating near-clinical testing of novel therapeutics.
Takahiro Morito¹, Naoto Watamura¹˒², Hiroki Sasaguri¹, Taisuke Tomita³, Makoto Higuchi⁴, Hideyuki Okano¹˒⁵, Erika Sasaki⁶, and Takaomi C. Saido¹*
¹RIKEN Center for Brain Science, Hirosawa, Wako, Japan
2UK Dementia Research Institute, University College London, London, United Kingdom
3Graduate School of Pharmaceutical Science, University of Tokyo, Hongo, Tokyo, Japan
4Institute for Quantum Medical Science, National Institutes for Quantum Science and Technology, Chiba, Japan
5Department of Physiology, School of Medicine, Keio University, Tokyo, Japan
6Central Institute for Experimental Animals, Kawasaki, Kanagawa, Japan
*Correspondence: takaomi.saido@riken.jp
References
[1] Morito T, Watamura N, Sasaguri H, et al. Experimental basis for generating nonhuman primate models of frontotemporal dementia and Alzheimer’s disease. J Alzheimers Dis 2025; DOI: 10.1177/13872877251321116.
[2] Huhe H, Shapley SM, Duong DM, et al. Marmosets as model systems for the study of Alzheimer's disease and related dementias: Substantiation of physiological tau 3R and 4R isoform expression and phosphorylation. Alzheimers Dement 2025; 21: e14366.
[3] Sharma G, Huo A, Kimura T, et al. Tau isoform expression and phosphorylation in marmoset brains. J Biol Chem 2019; 294: 11433-11444.
[4] Watamura N, Foiani MS, Bez S, et al. In vivo hyperphosphorylation of tau is associated with synaptic loss and behavioral abnormalities in the absence of tau seeds. Nat Neurosci 2025; 28: 293-307.
[5] Bellenguez C, Küçükali F, Jansen IE, et al. New insights into the genetic etiology of Alzheimer's disease and related dementias. Nat Genet 2022; 54: 412-436.
[6] UniProt Consortium. UniProt: the Universal Protein Knowledgebase in 2025. Nucleic Acids Res 2025; 53: D609-D617.
Dear Dr. Mellar P. Davis,
Thank you for carefully reviewing our manuscript titled "Rapid Cognitive Assessment: Accuracy and Discriminant Validity of Mini-Cog and Process-Based Clock Drawing Test" published in JAD and for pointing out the miscalculations in the negative likelihood ratio (NLR) values. We are grateful for your expertise and the time you spent examining our work.
Upon reanalysis of our original data, we confirmed that the NLR values in Table 3 were indeed calculated incorrectly, with the numerator and denominator reversed. We deeply regret this oversight, which may have mislead readers. Enclosed, please find the corrected Table 3, with the revised NLR values highlighted in red. The corrected NLR for the Clock Drawing Test (CDT) in diagnosing MCI is 0.21, and for the Mini-Cog3, it is 0.24. The small difference of 0.03 indicates that both tests have similar performance in ruling out MCI. However, the Mini-Cog3 shows a significantly higher area under the curve (AUC) of 0.82 compared to the CDT's AUC of 0.77, demonstrating its superior diagnostic accuracy.
Regarding your suggestion that “in the sequence of testing, the CDT should be administered first, and if the CDT is normal, further testing with the Mini-Cog for mild cognitive impairment may not be necessary,” we would like to clarify. While the CDT can indeed rule out MCI, if the CDT is abnormal, the Mini-Cog3 should not be administered subsequently, as it already includes the CDT. The Mini-Cog3 assessment begins with immediate recall of three words, followed by the CDT, and concludes with delayed recall of the same words. This comprehensive approach, which evaluates both memory recall and executive functions, is completed within a brief 2-3 minutes, making it an efficient tool for clinical settings. Consequently, we advocate for the direct use of the Mini-Cog3 in community screenings for MCI and early-stage Alzheimer’s disease to enhance detection rates without incurring significant time costs.
Given the minimal difference in NLR values and the superior diagnostic accuracy and comprehensive assessment provided by the Mini-Cog3, we maintain our recommendation that the Mini-Cog3 is the preferred tool for the rapid and accurate detection of cognitive dysfunction in patients with MCI and Alzheimer's disease.
Once again, we thank you for your insightful feedback.
Sincerely,
Qian Yang
Corresponding author: Defa Zhu, Department of Geriatric Endocrinology, The First Affiliated Hospital of Anhui Medical University, Hefei, Anhui, China. Email: zdfa0168@sina.com
Qihao Guo, Department of Gerontology, Shanghai Jiao Tong University Affiliated Sixth People’s Hospital, Shanghai, China. Email:
The corrected Table 3 is as follows:
Thank you sincerely for taking the time to read and acknowledge our research. We are also grateful for your insightful guidance regarding the shortcomings of our study. We have thoroughly reviewed your letter and have reflected deeply on the issues you highlighted in the text.
The ADNI database serves as a global specialized cohort for clinical research on Alzheimer's disease, offering a rich array of clinical and biomarker data. The project aimed to focus on the AD pathology and AD clinical performance, which could introduce selection bias and limit the generalizability to other clinical populations. Therefore, the next step is to validate our findings through studies that include more diverse populations. We whole heartedly support the author's proposal for standardizing ATN cutoff values. However, achieving this goal will necessitate collaborative efforts and extensive validation across multiple studies. We eagerly anticipate a future where standardized cutoff values for AD-related biomarkers are more widely available. Due to the absence of a frailty assessment program within the ADNI cohort and the lack of a standardized method for evaluating frailty, we have faced significant challenges in our assessment efforts. To address this, we have turned to the mFI-11 scale as applied by Soon et al. [1] in the ADNI cohort. We are optimistic about achieving a precise evaluation of frailty in this cohort in the future and plan to explore the replication of our findings using various assessment scales.
We sincerely appreciate your valuable feedback on our research. Your insights will guide us in enhancing our study in future. Thank you once again.
Bao-Lin Han1, Hui-Fu Wang1,2,*, Lan Tan1,2,*
1Department of Neurology, Qingdao Municipal Hospital, Qingdao University, Qingdao, China
2Department of Neurology, Qingdao Hospital, University of Health and Rehabilitation Sciences (Qingdao Municipal Hospital), Qingdao, China
*wanghuifu2010@126.com, dr.tanlan@163.com
References
[1] Soon SXY, Kumar AA, Tan AJL, Lo YT, Lock C, Kumar S, Kwok J, Keong NC (2021) The impact of multimorbidity burden, frailty risk scoring, and 3-directional morphological indices vs. testing for CSF responsiveness in normal pressure hydrocephalus. Front Neurosci 15, 751145.
A very good review about the mitochondrial cascade hypothesis of AD, including a series of aspects.
The investigators did an excellent job studying a 40-year cohort to provide definitive population evidence that oral health is a risk factor for dementia.
It is a research that builds evidence for new perspectives.