Congratulations to Xie Liwei, the biological research and development consultant, for securing the new bid for the Olympics

Globally, the prevalence of overweight/obesity is increasing rapidly. Obesity and its complications not only seriously affect the quality of life of patients but also bring a heavy economic burden to society and families. Low carbohydrate diets (LCD) are a dietary intervention mode for weight loss therapy. However, in different studies, the weight loss effects of LCD intervention are quite different. There is currently no sufficient evidence to explain this difference. This is a qualitative phenomenon, which is also a challenging aspect in the field of medical weight management.

On September 15, the teams of Professor Hong Chen and Professor Sun Jia from the Department of Endocrinology and Metabolism at Zhujiang Hospital of Southern Medical University, and Professor Xie Liwei from the Gut Microbiology and Health Team of the Guangdong Institute of Microbiology, published the clinical research report titled 'Gut microbiota serves as a predictable outcome of short-term low-carbohydrate diet (LCD) intervention for patients with obesity' in the journal Microbiology Spectrum. This study reported for the first time that the baseline characteristic of the intestinal flora is a determinant of the short-term low-carbohydrate diet (LCD) weight reduction effect in the overweight and obese population. The study builds an Artificial Neural Networks (ANN) model based on the baseline characteristics of the intestinal flora to predict the weight loss effect of LCD. The findings provide a new approach for clinical medical weight management and intervention strategies.

Globally, the prevalence of overweight/obesity is increasing rapidly. Since 1980, the prevalence of obesity in more than 70 countries has doubled. The population affected by obesity or obesity-related chronic metabolic diseases has increased to more than 2 billion [1]. According to data from the National Center for Health Statistics (NCHS), from 2017 to 2018, the prevalence of obesity in the U.S. was about 42.4%, and the prevalence of severe obesity with a BMI≥40 kg/m2 reached 9.2% [2]. At the same time, the "Report on Nutrition and Chronic Disease Status of Chinese Residents (2020)" [3] pointed out that the prevalence/incidence rate of overweight and obesity among Chinese residents is still rising rapidly, and the adult population's overweight or obesity rate has exceeded 50%. Overweight/obesity is a risk factor for a series of chronic diseases such as cardiovascular disease, type 2 diabetes, cancer, etc. [4], [5], which seriously endangers the health of Chinese people [6], [7], [8]. Additionally, there are more than 29 complications such as hypertension, dyslipidemia, and glucose metabolism disorders caused by obesity in adolescents, seriously affecting the physical development and health of adolescents. For obese patients, CVDs are the main reason for the high obesity-related mortality and disability rate. The high BMI-related disability rate caused by CVDs is 34%, and the high BMI-related mortality rate is as high as 41% [9]

Increasing morbidity, potential health hazards, and significant economic burden have made the problem of overweight/obesity a huge challenge in the field of global public health. In recent years, various forms of weight loss interventions have been gradually applied in clinical practice and written into guidelines. Lifestyle interventions are the cornerstone of obesity treatment, and dietary interventions are the primary choice. Among the many dietary intervention models, low-carbohydrate diet intervention has attracted much attention. It has a long history, but it has different forms. In recent years, LCD has attracted widespread attention, but there are also certain controversies.

This study included 51 male or female subjects aged 18-65 who met the diagnostic criteria of overweight/obesity (no antibiotics or drugs were used in the first 3 months of the clinical trial). The subjects were randomized into groups and were divided into different groups. Energy-restricted normal diet (ND) group and non-calorie-restricted low-carbohydrate diet group (LCD). The diet intervention time was 12 weeks. In order to ensure the LCD diet structure, the LCD group adopted a standardized nutrition bar (provided by Guangzhou Nanda Feite Nutrition and Health Consulting Co., Ltd.) instead of the daily staple food for lunch and dinner. The number of other foods is not limited, and overeating is avoided. At the time of enrollment (ie baseline) and 12 weeks after the intervention, venous blood and stool samples were collected. The blood samples were used for the detection of blood biochemical indicators such as glucose and lipid metabolism, liver and kidney function, and the stool samples were used for intestinal flora 16S rDNA amplicon sequencing, through 16S rDNA amplicon sequencing, a total of 2.47 million high-quality reads were obtained. The diet of the subjects was monitored through a 24-hour diet for 3 days a week. During the entire study period, the average proportion of carbohydrate intake in the normal diet group was about 50%, and the proportion in the LCD group was about 20% (Figure 1B-D). Although calorie intake was not restricted, the average energy intake of the low-carbon group was about 50%. Enrollment was significantly lower than that of the normal diet group. The 12-week LCD intervention significantly improved the subjects` body parameters such as BMI, waist circumference, waist circumference, body fat percentage, and visceral fat area.

In addition to different weight loss results, different dietary components may affect the composition and diversity of the intestinal flora, but apart from changes in the overall composition and phyla level, previous studies have not drawn a constructive conclusion to guide LCD Therefore, we have analyzed the intestinal flora sequencing data, and adopted 5-fold cross-validation and random forest algorithm, taking into account the minimum error rate and standard deviation to ensure the highest accuracy and stability. Next, we analyzed the 16S rDNA sequence data of subjects in the ND and LCD groups before and after the test to identify potential biomarkers of flora. A further analysis of all the genus screened by the random forest model in baseline and week 12 data found that the relative abundance of Ruminococcaceae Oscillospira and Porphyromonadaceae Parabacteroides increased significantly after the 12-week LCD intervention, and the difference was statistically significant (p< 0.05). According to existing research reports, these two species of bacteria are involved in the production of butyrate in the intestine, suggesting that there may be other factors affecting weight changes during the process of LCD intervention in weight loss.

Further analyze the weight loss of each subject, and divide each group into two subgroups according to the clustering of weight loss parameters: BMI, waist circumference, WHR, BFR and VFA changes: a moderated weight loss effect. weight loss group (MG) and significant weight loss group (distinct weight loss group, DG). Under the conditions of LCD intervention, the energy intake and the proportion of carbohydrates in the diet of the two subgroups were almost the same, but the weight loss indexes of the subjects in the markedly effective subgroup decreased more significantly, suggesting that the individualized differences in weight loss effects may be affected by other factors. Influence

The above results suggest that LCD intervention has a good weight loss effect, but there are individual differences. Therefore, this study further analyzed the intestinal flora data of the two subgroups, and further explored whether there are potential factors related to the flora that caused the difference in weight loss between the two subgroups in this diet. In further subgroup analysis, we used the co-occurrence network at the genus level to further analyze the interaction between the intestinal flora in the LCD subgroup and found that after 12 weeks of LCD intervention, although the two subgroups LCD_DG and LCD_MG The network interaction complexity of the network has decreased, but LCD_DG showed a denser, more extensive and richer network interaction complexity than LCD_MG in the baseline and at the 12th week. The above results indicate that, in addition to the differences in the composition and diversity of the flora, the differences between the structure of the flora and the complexity of the interaction of the flora may be an important reason for the individual differences in the weight loss effect. In the low-carbon subgroup, analysis by the random forest model algorithm found that the baseline relative abundance of Bacteroidaceae Bacteroides was statistically different in the two subgroups of the low-carbon diet. According to linear regression analysis, we found the baseline of Bacteroidaceae Relative abundance is positively correlated with the weight loss effect of short-term low-carbon diet. Based on the above results, the ROC model was established based on the baseline relative abundance of the low-carbon subgroup of Bacteroides. The ROC model reflects the susceptibility of each data point on the curve to the same signal stimulus and comprehensively reflects the sensitivity and specificity of the variables. In this study, the ROC model AUC value reached 73.2%, suggesting that the baseline relative abundance of Bacteroides has a certain predictive value for the short-term low-carbohydrate diet weight loss effect.

Since the flora in the human intestine is not an independent individual, there are intricate connections between bacteria. Therefore, this research introduces artificial neural network (ANN). ANN is a more powerful deep learning model that is trained and used to simulate biological neural networks for complex data analysis. ANN is based on biological neural networks. The basic principle of the network, imitating the human brain structure and external stimulus response mechanism, building a model based on the knowledge of network topology, has the functions of associative memory, classification and recognition, optimized calculation and nonlinear mapping. In recent years, more and more medical researches apply ANN to the processing of complex data. We incorporated the change values and ratios of the weight loss parameters of the LCD group into the ANN model based on the baseline relative abundance of the overall intestinal flora of the group, and obtained a higher predictive model determination coefficient (R2), which also indicates the prediction of ANN The effect is better than the linear model, suggesting that the prediction effect is better.

All in all, the current research shows that in overweight/obese people, short-term LCD intervention without calorie restriction has a significant weight loss effect without significant adverse effects. There are individual differences in short-term LCD weight loss. The relative abundance of Bacteroidaceae Bacteroides at baseline before LCD intervention is positively correlated with the short-term LCD intervention weight loss effect. Finally, this study constructed a high-precision ANN prediction model based on the relative abundance of the intestinal flora at the baseline. Through the ANN prediction model, it was found that the baseline relative abundance of the intestinal flora can be used as a predictor of the individualized weight loss effect before LCD intervention. , It has important guiding significance for clinical medicine weight management. Related research results were published in "Microbiology Spectrum".

Based on the results of this research, in clinical medicine weight management, the relative abundance of Bacteroidaceae Bacteroides in the intestines is relatively low, but the overweight/obese subjects who hope to lose weight through LCD may increase the weight loss of LCD by supplementing the corresponding Probiotics. Efficacy. At present, our research group is working with the Guangdong Academy of Sciences Institute of Microbiology and Xie Liwei Research Institute team to carry out clinical weight loss research on the combined use of Probiotics and low-carbon diet to further explore the strategies and ideas of medical weight management. Let us look forward to the updated research Results.

The main author of this study is Zhang Susu, a physician in the Department of Endocrinology and Metabolism, Zhujiang Hospital, Southern Medical University; Co-first author, Wu Peili, PhD candidate in the Department of Endocrinology and Metabolism, Nanfang Hospital, Southern Medical University; Tian is also a Ph.D., Department of Endocrinology and Metabolism, Zhujiang Hospital, Southern Medical University, Chen Hong The professor and researcher Xie Liwei of the State Key Laboratory of Applied Microbiology in South China jointly cultivated master students; Liu Bingdong is a joint training of PhD students by Professor Pan Jiyang from the Department of Psychiatry of the First Affiliated Hospital of Jinan University and researcher Xie Liwei of the State Key Laboratory of Applied Microbiology in South China. The corresponding author of this article is Professor Sun Jia from the Department of Endocrinology and Metabolism, Zhujiang Hospital of Southern Medical University, and the co-corresponding authors are Professor Chen Hong from the Department of Endocrinology and Metabolism, Zhujiang Hospital of Southern Medical University, and researcher PI Xie Liwei from the Intestinal Microecology and Health Team of the Institute of Microbiology, Guangdong Academy of Sciences.