Relationship between traditional maternal diet pattern and breastmilk composition of rural lactating women during the first month postpartum in Shigatse, Tibet

Abstract Maternal nutrition can influence the composition of human breastmilk by altering the components that are sensitive to maternal diet pattern. Traditional Tibetan maternal diet pattern among native rural lactating women possesses distinct characteristics due to its unique geographical environment and dietary habits. This study investigated maternal diet pattern and human breastmilk composition of Tibetan lactating women through different lactation stages during the first month postpartum in Shigatse. The results indicated that Tibetan maternal diet profile was apparently monotonous, mainly sufficient in Zanba, buttered tea, red meat, and fatty soup, yet insufficient in white meat, eggs, leafy vegetables, and fruits, leading to imperfect maternal nutritional intakes with high‐level carbohydrates and deficient proteins. Distinctions of maternal diet profiles in various degrees can be discovered upon different lactation stages, which brings multiple influences to the composition of human milk. There was significantly weak‐to‐medium correlation of protein contents between maternal diet intakes and human milk, while other macronutrients correlated insignificantly. Micronutrient constituents in human milk, involving functional unsaturated fatty acids and free essential amino acids, were also impacted by maternal diet intakes through different lactation stages. These results show that more systematic and profound research is requisite for the clarification and development of Tibetan maternal diet to offer more enhanced and individualized nutritional recommendations for Tibetan lactating women and infants.


| INTRODUC TI ON
Human breastmilk has always been the most idealistic natural nutritional source to meet the developmental needs of infants (Kramer & Kakuma, 2004). World Health Organization (WHO) has encouraged lactating women to breastfeed exclusively at the first 6 months to adequately ensure infants' multiple nutrients and bioactive components (Andreas et al., 2015;santé et al., 2003). The composition of human milk is a complex and dynamic system, which is affected by various internal and external elements. Among the influencing factors, maternal diet pattern has been stated to play an important role on the variation of breastmilk composition, especially on the essential components that are hard to be synthesized by human own metabolism, such as essential fatty acids and amino acids (Ding et al., 2010;Kim et al., 2017).
Macronutrients, that is, fats, proteins, and carbohydrates, constituent the primary composition of human milk, which are the most abundant nutritive portion and inferior to the proportion of water in human milk . Thereinto, milk fat is the main source of energy for infants. Meanwhile, milk fat also provides functional polyunsaturated fatty acids (omega-3 and omega-6 PUFA) and lipid-soluble vitamins (A, D, E, K), as well as influences the taste and mouthfeel quality of human milk (Koletzko et al., 2011). Moreover, the content and composition of milk fat are more variable than some other human milk components (such as carbohydrate), comparatively more sensitive to internal and external factors from mothers, infants, or lactation stages (Demmelmair & Koletzko, 2018). Maternal diet has been indicated to be correlated with the fat composition in human milk in previous publications, especially regarding the association of functional PUFA between human milk and maternal diet (Aumeistere et al., 2018;Jiang et al., 2016;Visentin et al., 2019;Yuhas et al., 2006). On the other hand, free amino acids (FAAs) were an important moiety of nonprotein nitrogen fraction in human milk, which promote the utmost utilization of dietary nitrogen and the regulation effect on infant early postnatal development due to the higher absorption efficiency of FAAs than protein-derived amino acids (Agostoni et al., 2000;Zhang et al., 2013). In particular, certain specific amino acids from FAA in human milk can exert more impact on the initialization of relevant biological functions during infancy, involving protein synthesis, appetite control, and growth regulation (Agostoni et al., 2000;Elmastas et al., 2008). The influence of maternal diet on the level of nitrogen fraction in human milk has been discussed earlier than the 1970s with controversial viewpoints, especially on the association of protein contents between maternal diet and human milk (Boniglia et al., 2003;Wurtman & Fernstrom, 1979).
Albeit such consequences, some nitrogen proportions in human milk (e.g., essential amino acids, whey-to-casein ratios, certain functional proteins) were considered to be correlated with maternal nutritional intakes, which could be more obvious among malnutrition women (Boniglia et al., 2003).
Tibet of China, noted as "the third pole of the world," lies in the southwest of the Tibetan Plateau with an average altitude of over 4,500 m, which is the highest plateau of the world (Xie et al., 2014;Yang et al., 2016). Because of its peculiar geographical location, formidable climatic conditions, and unique traditional customs, the local maternal and child nutritive style possesses distinct characteristics in various aspects. However, the exploration and development of relevant maternal and child health research has remained limited, especially on the maternal breastfeeding and infant development, which should be due to the inconvenience of resource utilization and mutual communication by the extreme natural conditions. Shigatse, the second largest city of Tibet, locates between the middle part of the Himalayas and the middle part of the Gangdise-Nyenchen Tanglha Mountains, within which stands Mount Everest (Yang et al., 2016). The meaning of "Shigatse" in Tibetan language is a great farm with the best soil, indicating its advantage of agricultural output (Yang et al., 2016). Meanwhile, Shigatse has a population of about 850,000, 89.9% of which is Tibetan and over 80% is farmers and herdsmen. Hence, the diet pattern of local lactating women is presumably capable to present the traditional Tibetan maternal dietary feature in nowadays.
Through the investigation of Tibetan maternal diet habit, the anthropometric measurement of local lactating women, and the analysis of human milk components during the first month postpartum were performed via prospective single-center cohort study.
The aim of this research was in an effort to explore the association between the Tibetan maternal diet pattern and the human milk composition, which hoped to be conducive to the reveal of maternal breastfeeding status and the development of local maternal and infant nutritional level at high-altitude areas with traditional culture.

| Participant recruitment
The prospective single-center cohort study was conducted from June 2018 to February 2019, which had acquired prior approval from the Ethics Committees of the Chinese Capital Institute of Pediatrics (SHERLL2014034) according to Helsinki declarations involving human subjects. At the beginning of the study, participants were recruited through random invitations at the local maternal and child clinics in Shigatse. After individual assessment and face-to-face interviews, 50 healthy (free of chronic/genetic/infectious diseases) expectant mothers, who were largely anticipated to deliver healthy term infants, were chosen from the local Tibetan applicants in various rural sections of Shigatse. Emphatically, the recruited mothers were eligible for the criteria of native residency in Shigatse, restriction on Tibetan diet tradition, and individual will and capacity in breastfeeding.

| Questionnaire
With assistance from our trained and experienced researchers, basic information and dietary items were obtained from the recruited mothers during every sampling period on personal sociodemographic characteristics, physical status, dietary intakes, medical histories, and other necessities. Explicitly, maternal relevant data were recorded at the following time points: (a) necessary personal information acquired at postpartum periods among 0-5 days (colostrum stage), 7-15 days (transitional milk stage), and 26-30 days (mature milk stage) when breastmilk samples were collected; and (b) detailed dietary intakes recorded for 3 days before the collection of breastmilk samples. A 32-diet-item food frequency questionnaire (FFQ) with practicable validity and reproducibility was adopted for the diet investigation of local lactating women. The items of food products were divided into 12 groups according to previous relevant reports and local Tibetan diet habits in Shigatse (Cheng et al., 2009;Z. Wang et al., 2010). The feedback of FFQ was consisted of the frequency and consumption quantity of a certain food or beverage, and its existing form in certain relevant recipe. The evaluation of food intakes by FFQ was aimed to assess the level of nutrition intakes of local lactating women with adherence to traditional Tibetan diet pattern. Macronutrient intakes of carbohydrate, protein, fat, and energy were calculated according to the 2017 China Food Composition Tables (INFS, C . C., 2017), corresponding to the items of macronutrients in human milk analysis.

| Breastmilk sample collection
After finishing 3 × 24 hr daily diet questionnaires, participants were asked to collect complete either-side breastmilk by electric breast pump in the next forenoon between 09:00 a.m. and 11:00 a.m. in a sanitary and quiet environment. 60-ml aliquot of the well-shaken breastmilk was then divided equally into 10-ml sterilized tubes. All the breastmilk samples were prestored at −20°C for no more than 7 days before being transported by ice box to −80°C refrigerators until analysis.

| Macronutrient content analysis
The macronutrients of fat, crude protein, lactose, total solids, true protein, and energy in human milk samples were measured by human milk analyzer (MIRIS HMA, MIRIS AB, Uppsala, Sweden) with XMA-SW software version 2.87. MIRIS HMA determined the macronutrient contents of human milk based on semi-solid mid-infrared transmission spectroscopy under different waveband conditions, which can be specifically as follows: 5.7 μm for functional carbonyl groups (fat content determination), 6.5 μm for amide groups (protein content determination), and 9.6 μm for hydroxyl groups (carbohydrate content determination). According to the recommended procedure from the manufacturer, human milk samples, which were stored at −80℃ before analysis, should be homogenized for 30 s by ultrasonic probe (MIRIS Sonicator, MIRIS AB) after being preheated to 40°C. Prior to analysis, preparation steps of self-calibration check and autocleaning check were performed between every continuous 10th tests or without measurement within 5 min. Prepared human milk sample (3 ml) was injected into the flow cell and started to test immediately. Within 1 min, the test data would be shown on the system software.

| Fatty acid analysis
The composition of fatty acids in human milk was analyzed by gas chromatography-mass spectrometry (GC-MS) with the pretreatment of n-hexane extraction and transesterification by methanol. Milk sample (250 μl) was mixed with 5 ml HCl-methanol (1:1, 0.5 mol/L) solution, 2 ml n-hexane, and 2 ml methanol. The mixture was incubated at 80°C for 2 hr with mechanical stirring, which was cooled to room temperature by flowing water afterward. With the addition of 2 ml deionized water, the mixture was centrifuged at 2352 g for 5 min. Finally, 1 ml of liquid supernatant was extracted for GC-MS analysis. Fatty acid methyl esters (FAMEs) were detected using GC-MS (Thermo Trace 1300-Thermo ISQ LT, Thermo Scientific) with an autosampler (Thermo Scientific TriPlus RSH).

μl of FAMEs was injected at 200°C into a HP-88 capillary column
(100 m × 0.250 mm × 0.20 μm) at a 10:1 split ratio. Nitrogen was used as carrier gas at the flow rate of 1 ml/min. The oven temperature program was set as follows: the initial time was programmed at 60°C and held for 5 min, then increased at 8°C/min to 160°C, next increased at 4°C/min to 200°C and held for 5 min, and finally increased at 3°C/min to 240°C and held for 5 min. The solvent delay was 10 min. The temperature of ion source and interface was

| Free amino acid analysis
The composition of free amino acids in human milk was analyzed by ultraperformance liquid chromatography (UPLC) with photo-diode array (PDA) detector using AccQ·tag Ultra C18 1.7 μm, 2.1 × 100 mm column, and AccQ-Fluor Reagent Kit (all from Waters Corporation) for precolumn derivatization. Milk sample (300 μl) was mixed with 20 μl 10% sulfosalicylic acid in a 1.5-ml tube, which was then centrifuged for 15 min at 9408 g and 4°C and filtered through 0.22μm ultrafiltration membrane. 10 μl filtrate was mixed with 70 μl borate buffer (0.4 M) and 20 μl AccQ-Fluor in a UPLC tube, which was incubated at 55°C for 10 min and then cooled to ambient temperature for UPLC analysis. The same method was applied for the standard solutions (2.5 μmol/ml) of 17 amino acids, which was histidine, serine, arginine, etc. 1 μl of derivatized samples was injected into the UPLC system, which was detected by PDA detector at 260 nm.
The solvent used for mobile stage was the manufacturer-supplied AccQ-Tag Ultra Eluent A and AccQ-Tag Ultra Eluent B. The UPLC gradient program (A = AccQ-Tag Ultra Eluent A, B = 10%AccQ-Tag Ultra Eluent B, C = water, and D = AccQ-Tag Ultra Eluent B) is listed in Table 1.

| Maternal and infant characteristics in this cohort study
As shown in Table 2, 45 (45/50) expectant mothers managed to complete the thorough participation of this cohort study with enough supply of effective data and breastmilk samples. Specifically, 44 expectant mothers had healthy term infants and 1 had preterm infants, which were consisted of 21 baby boys and 24 baby girls. Meanwhile, other 5 (5/50) mothers failed to finish the participation for the reasons of miscarriage, incapable to supply complete breastmilk and data or other individual matters. The participants had a mean age of 27.21 years (SD = 4.48), which was fell into the range of prime ages for childbearing (Thalberg, 2011). The mean prepregnancy BMI was 21.66 kg/m 2 (SD = 2.20). 35 participants had managed to maintain fit and fine for pregnancy. According to our detailed records, the main symptoms of other 10 participants during pregnancy were mild anemia and muscle cramp, which might be due to some malnutrition causes, such as iron or calcium deficiency. Before or during the pregnancy period, 37 participants used nutritive supplements, most of which were folic acid, iron supplement, or blood-supplementing formula. Prior to our study criteria, the occupations of 46 recruited pregnant women were local Tibetan farmers and herdsmen, for a better assurance in reflecting traditional Tibetan diet pattern. 44 infants were delivered by natural vaginal mode, and 1 was through cesarean section. The mean birthweight and birth length of infants were 2,984.4 g (SD = 418.6) and 48.4 cm (SD = 4.0), respectively.

| Characteristics of Tibetan diet pattern among lactating women in Shigatse
The description of diet pattern among Tibetan lactating women in Shigatse is revealed in Table 3. Summarily, the selection of certain food types has basically possessed a shared commonness by Tibetan diet tradition, with some varied characteristics by individual maternal preference during the lactation period. Specifically speaking in food categories, Zanba (a traditional Tibetan food, which is cooked by roasting highland barley flour, usually mixed with buttered tea or red meat broth (Qu et al., 2015)) was the dominating cereal type for staple food, consumed by over 98% participants through all three lactation stages. Other cereal choices included noodle, rice, and deep-fried dough stick. Particularly, Zanba seemed to be an exclusive choice of cereal among lactating women in colostrum stage, while the profile of cereal category would exhibit more diversity in the following transitional milk and mature milk stages. Moreover, Zanba was also used as the first type of cereal introduced to local Tibetan infants in about 1 month after birth according to previous publications (Dang et al., 2005). Such priority to Zanba in local maternal and infant diet pattern should be owed to the long-held tra-  presumably linked with the potential adverse outcomes if manageable breastfeeding strategies can be adopted (Wilson et al., 2017).
Red meat, generally including beef and mutton, were the mainly consumed meat among the participants, which were usually cooked with Zanba and buttered tea or other traditional Tibetan staple foods, while white meat, such as poultry meat and fish, was rarely involved in their daily diet. Such inclination to red meat rather than white meat might be due to the high-altitude environment and traditional Tibetan diet culture. Moreover, eggs were not a commonly involved food category in the participants' recipe, possibly on the account of less adoption upon poultry farming than livestock farming in such areas (Shang et al., 2014). It is unsurprisingly that pure milk was not taken as a drink for the lactating women, since milk was conventionally used as the ingredient for traditional Tibetan drinks and beverages, such as sweet tea. Vegetables and fruits were not a daily acquisition for every lactating mother, especially in the colostrum stage. Additionally, the ordinary vegetables were root and tuber vegetables, such as turnips, carrots, tomatoes, and potatoes, while leafy vegetables and fruits were taken less frequently, which were generally spinach, Chinese cabbage, pakchoi, and apple and banana according to our detailed information.
Consequently, the lactating women in the rural areas of Shigatse were customarily prone to a traditional Tibetan dietary pattern, invariably including Zanba, Tibetan buttered tea and sweet tea, red meat, fatty soup, and root and tuber vegetables with the insufficiency of leafy vegetables, fruits, and white meat, which was in accordance with the findings of previous relevant reports (Ge et al., 1997;Wang et al., 2010

| Contents of macronutrients in human milk from Shigatse
The macronutrient contents of fat, crude protein, lactose, total solids, true protein, and energy in human milk are shown in Table 4

| Composition of fatty acids in human milk from Shigatse
The fatty acid composition of human milk from Shigatse is displayed in The expression of "every day," "sometimes," or "never" means a certain food was taken every day, no less than once and no more than every day, or never taken by participants during the 72-hr sampling time.

| Composition of free amino acids in human milk from Shigatse
As described in

| Classification of maternal diet profiles through different lactation stages during the first month postpartum
Principal component analysis (PCA) was adopted to display the classification of maternal diet profiles through three lactation stages.

| Correlation analysis of macronutrients between maternal diet profiles and human milk during different lactation stages
As shown in Note: Values are means ± SD.
rural areas in western China (Cheng et al., 2009;Liu et al., 2015;Wang et al., 2010). Moreover, the intakes of energy and fat were through similar variation trends during the three lactation stages: initially held at significantly low (p < .05) levels in the colostrum stage, then increased significantly (p < .05) in the transitional milk stage and decreased significantly (p < .05) in the following mature milk stage. Note: Values are means ± SD.
Abbreviations: EAAs, essential amino acids, including His, Leu, Lys, Phe, Val, Thr, Met, Ile, and Tau; TAA, total amino acid. a Different lowercase letters in a row represent statistical differences among samples, p < .05.
F I G U R E 1 PCA and correlation analysis of stage differences in the food profile upon traditional Tibetan maternal diet pattern. (a) PCA score plot of food profiles in different lactation stages. (b) Correlation loadings of food products for PCA classification Furthermore, the intakes of protein were at low levels with insignificantly varying degrees (p > .05) through the lactation stages, while the intakes of carbohydrate increased significantly (p < .05) from colostrum stage to transitional milk stage and kept at a relatively steady level (p > .05) afterward. Correlated with the diet frequency of food products among participants (Table 6) and their living style, the above results of daily macronutrient intakes can be interpreted as follows: (a) dietary patterns-according to the Tibetan dietary habit, the main type of food products was the staple food, basically cereals, among which highland barley was at a dominating dietary level. As is known, cereals are abundant in carbohydrate and energy rather than in protein and fat (e.g., highland barley in average: 62 g/100 g in carbohydrate and 298 kcal/100 g in energy, but 10.2 g/100 g in protein and 1.2 g/100 g in fat; rice in average: 76 g/100 g in carbohydrate and 352 kcal/100 g in energy, but 12 g/100 g in protein and 0.8 g/100 g in fat) (INFS, C. C., 2017), which would thus lead to a relatively more carbohydrate and less protein intake profiles in their daily diet; and (b) lactation stages. According to the findings of dietary frequency survey, the diet options were varying between either lactation stages.
Explicitly, lactating women in general were prone to a monotonous diet pattern in colostrum stage, merely including Zanba, buttered tea, and red meat and bone soup, which could be reflected in the shortage of daily energy, protein, and fat intakes. The diet profiles favored relatively more diversified combination in transitional milk stage, with more occurrence of other cereals, eggs, leafy vegetables, and fruits, which significantly (p < .05) enhanced the intake values of energy, fat, and carbohydrate. In mature milk stage, the selection of food species was further more various and individualized in each food category, while such variability was confined to the substitution within the same food category, which did not present obvious differences across different food categories. Moreover, the intake values of energy and fat in mature milk stage were significantly lower than the intakes in transitional milk stage, which might be due to the alteration of maternal nutritional needs since the extended duration of lactation period.
Bivariate correlation analysis was processed upon a homogeneous comparison of macronutrients between daily maternal diet intakes and the corresponding human milk samples by SPSS. Interpretively, there are mainly two constituent parts of carbohydrates in human milk, namely lactose and oligosaccharides. Thereinto, lactose consists the principal proportion (generally over 90%) in human milk to ensure appropriate nutrition and development needs for infants, while oligosaccharides in human milk, primarily as prebiotics, exert different effects in the infant gut (Gridneva et al., 2019). Therefore, it is admissible to correlate lactose in human milk with carbohydrates in maternal diets upon the aspect of macronutrients.
As displayed in Table 7, only the item of protein displayed varying significant degrees of correlation through the three lactation stages, whereas other macronutrients in human milk were uncorrelated with the relevant macronutrients in daily intakes. Such results were not quite parallel with some previous relevant reports (Aumeistere et al., 2019;da Cunha et al., 2005). In this research, the intake of protein was negatively weak-correlated between human milk and maternal diet in the colostrum stage, then positively medium-correlated TA B L E 7 Daily values of macronutrient and energy intakes from maternal diets and their correlation analysis with human milk during lactation  in the transitional milk stage, and finally uncorrelated in the mature milk stage. Meanwhile, protein intakes from maternal daily diet were lower than the values of Chinese RNIs through all the three lactation stages. However, the protein contents in human milk samples did not possess obvious inadequacy accordingly in the corresponding lactation stages (Table 3), which was consistent with some relevant publications (Shi et al., 2011;Zhu et al., 2017). Such consequence might reveal that protein shortage in daily maternal diet was still a restrictive factor for the lactating women, especially in the colostrum and transitional milk stages. Benignantly, the nutritional metabolism involving multiple constituents in maternal digestion tract and mammary gland would be conducive to the maintenance of protein content in human milk. Meanwhile, the overcapacity contents of carbohydrate in maternal daily diets did not induce the excessive proportions of carbohydrate in their human milk, which indicates that the internal metabolism might be more influential than the external maternal dietary profiles to the ultimate composition of human milk.

| Correlation analysis of key functional micronutrients between maternal diet profiles and human milk during different lactation stages
Cluster heatmap analysis was applied to show the differences and similarities of the key functional micronutrients in this study among human milk samples from three lactation stages, including free essential amino acids (His, Leu, Lys, Phe, Val, Thr, Met, Ile, and Tau) and main functional unsaturated fatty acids (LA, ALA, ARA, EPA, and DHA), as is shown in Figure 2. The content variations of the key functional micronutrients clustered into three categories. Cluster 1 was consisted of His, Leu, Lys, Phe, Val, Thr, Met, Ile, and ALA, ARA, EPA, and DHA, the contents of which generally decreased nonmonotonically according to orders of human milk samples; Cluster 2 was consisted of LA, the content of which increased nonmonotonically to a relatively high level accordingly; and Cluster 3 was consisted of Tau, the content of which decreased accordingly from a relatively high level. The cluster findings among the human milk samples indicated that the content variations of the involved micronutrients were not distinctly classified according to different lactation stages. However, some potential relevance could be implied when we divided the human milk samples into three groups (as is Group A, Group B, and Previous publications have considered that micronutrients, such as functional unsaturated fatty acids and free essential amino acids, might be more dependent on the daily dietary conditions during lactation (Aumeistere et al., 2018(Aumeistere et al., , 2019Garcia-Rodenas et al., 2016;Kim et al., 2017). Generally, the food categories of seafoods, eggs, and nuts are important sources for functional unsaturated fatty acids, while such kinds of food were not frequently enough prepared in the participants' daily diets. In consequence, the contents of main functional fatty acids in the human milk samples of this research, including PUFA, n−3 fatty acids, and n−6 fatty acids, were mostly insufficient than the published data from Chinese urban areas and other similar regions (Cruz-Hernandez et al., 2013;Jiang et al., 2016;Thakkar et al., 2019). Meanwhile, the supply of FAAs in human milk, especially essential FAAs, which likewise rely on the quality of maternal daily diets (Zhang et al., 2013), was also deficient among the participants in this research than in other more developed areas (Garcia-Rodenas et al., 2016;Liang et al., 2018).

| LI M ITATI O N S
This study has its limitations. First, although the effect of maternal diet pattern on human milk composition of macronutrients and micronutrients has been discussed in various studies, the impact of traditional maternal diet patterns on human milk composition at Tibetan high-altitude areas was poorly documented in recent publications, especially aiming at rural farmer groups. Therefore, few previous relevant reports are available for explicit analysis and clarification. Secondly, the participants in this study were scattered living across the rural areas of Shigatse with limited means of transportation, which brought more difficulty to collect samples and survey files. Consequently, the recruitment and follow-up survey of fifty effective participants were a realistic attempt to investigate the traditional Tibetan maternal diet pattern and its impact on the human milk samples during different lactation stages, while it would be more considerable if a wider range of cohort research with larger sample capacity was available in future.

| CON CLUS IONS
Traditional maternal diet pattern during the first month postpartum in rural areas of Shigatse, Tibet, was depicted in this research, which was monotonously abundant in Zanba, buttered tea, red meat, and fatty soup, yet inadequate of white meat, eggs, leafy vegetables, and fruits. Accordingly, the nutritional values of maternal diet intakes were high in contents of carbohydrate and deficient in contents of protein when compared to the values of Chinese RNIs. Maternal diet profiles presented distinguishable characteristics in various degrees though different lactation stages, which influenced the composition of human milk. Except for the weak-to-medium correlation of protein contents, the composition of maternal diet insignificantly affects other macronutrient levels of human milk during the lactation period. Conversely, micronutrient components, including functional unsaturated fatty acids and free essential amino acids, were impacted by maternal diet conditions, which were in insufficient levels than more developed regions in Chinese urban areas and other similar regions.
Thus, more efforts should be required to support the enhancement of local maternal nutritional level and the delicacy cultivation of maternal and infant health care.

ACK N OWLED G M ENTS
All the participants and volunteers in this study are heartily acknowledged for sample and data provision. Senior study nurse Yuzhen Ciren is acknowledged for the sample collection and data survey all along the cohort study. Guowen Wang and Tie Lin are appreciated for the sample transport from Shigatse to Beijing.

CO N FLI C T O F I NTE R E S T
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.