Table 2 Mendelian randomization studies in the bone field
Studies | Exposure | Outcome | IVs | Sample size; ethnicity | MR method | Unit | P value | Estimate (95% CI) | Interpretation |
|---|---|---|---|---|---|---|---|---|---|
Timpson NJ (2009)94 | Obesity (fat mass) | BMC | 2 SNPs (fat mass) | Obesity (8 480) BMC (7 470), children; European | IV regression with 2SLS | 1 g change in BMC per 1 kg change in fat mass | 0.000 2 | TB-BMC, 0.02 (−0.20, 0.15) | Fat mass is in the causal pathway for bone mass in children. |
0.03 | UL-BMC, 0.46 (0.31, 0.61) | ||||||||
0.002 | LL-BMC, 0.55 (0.41, 0.68) | ||||||||
2.30E-06 | LS-BMC, 0.48 (0.33, 0.63) | ||||||||
Warodomwichit D (2013)96 | Obesity | BMD | 1 SNP | Obesity (2 154) BMD (2 154), adults; Thai | IV regression with 2SLS | 1 g.cm−2 change in BMD per 1 kg.m−2 change in BMI | 0.01 | TH-BMD, 0.02 (0.00, 0.03) | Obesity might be causally related to BMD at the femur but not at the spine. |
0.014 | FN-BMD, 0.01 (0.00, 0.03) | ||||||||
NS | LS-BMD, 0.00 (−0.01, 0.01) | ||||||||
Oeil L (2014)176 | Inflammation | Fracture | 29 SNPs (CRP) | Inflammation (6 386) Fracture (1 561), adults; American | Weighted genetic risk score | OR for fracture per 1 SD increase in CRP | 0.23 | Fracture, 1.00 (0.99, 1.00) | No causal association between CRP level and fracture. |
Leong A (2014)84 | Serum DBP levels | BMD | 1 SNP | Serum DBP levels (2 254) BMD (2 254), adults; Canadian | IV regression with 2SLS | 1 g.cm−2 change in BMD per 1 SD change in DBP | 0.43 | FN-BMD, −0.005 (−0.02, 0.01) | No causal association between DBP level and BMD. |
Dalbeth N (2015)98 | Urate | BMD | 5 SNPs | Urate (2 501) BMD (2 501), adults; European | IV regression with 2SLS | 1 g.cm−2 change in BMD per 1 mmol.L−1 change in urea levels | 0.06 | TF-BMD, −0.29 (−0.60, 0.01) | No causal association between urate and BMD. |
0.08 | FN-BMD, −0.27 (−0.58, 0.03) | ||||||||
0.68 | LS-BMD, 0.08 (−0.32, 0.48) | ||||||||
Xiong A (2016)78 | Urate | BMD | 18 SNPs | Urate (1 322) BMD (1 322), adults; Chinese | IV regression with 2SLS | 1 g.cm−2 change in BMD per 1 mmol.L−1 change in urea levels | 0.5 | TH-BMD, 0.19 (−0.36, 0.74) | No causal association between urate and BMD. |
0.53 | FN-BMD, −0.19 (−0.42, 0.81) | ||||||||
0.26 | LS-BMD, 0.39 (−0.26, 0.98) | ||||||||
Kemp JP (2016)95 | Obesity | BMD | 32 SNPs (BMI) | Obesity (5 221) BMD (5 221), children; European | MR Egger; Multivariable MR | SD change in BMD per SD increase in BMI | 0.78 | SK-BMD, −0.02 (−0.20, 0.15) | Obesity is causally related to increase in BMD at all sites except the skull. |
<0.001 | UL-BMD, 0.46 (0.31, 0.61) | ||||||||
<0.001 | LL-BMD, 0.55 (0.41, 0.68) | ||||||||
<0.001 | LS-BMD, 0.48 (0.33, 0.63) | ||||||||
<0.001 | PE-BMD, 0.39 (0.34, 0.64) | ||||||||
Li SS (2016)85 | Vitamin D level | BMD | 4 SNPs | Vitamin D level (1 824) BMD (1 824), postmenopausal women; Chinese | IV regression with 2SLS | 1 g.cm−2 change in BMD per 1 log ng.mL−1 change in total 25OHD | 0.326 | TH-BMD, −0.04 (−0.13, 0.04) | No causal association between vitamin D and BMD. |
0.261 | FN-BMD, −0.04 (−0.13, 0.03) | ||||||||
0.384 | LS-BMD, 0.05 (−0.16, 0.06) | ||||||||
Ahmad OS (2017)89 | T2D | BMD | 32 SNPs | T2D (149 821) BMD (32 961), adults; European | IVW approach | SD change in BMD per odds in log-odds of T2D | 0.044 | FN-BMD, 0.034 (0.001, 0.067) | Genetically increased T2D risk and genetically increased fasting glucose have weak positive effects on FN-BMD. |
0.148 | LS-BMD, 0.026 (−0.01, 0.061) | ||||||||
Fasting glucose (FG) | BMD | 30 SNPs | FG (133 010) BMD (32 961), adults; European | IVW approach | SD change in BMD per 1 mmol.L−1 increase in GF | 0.034 | FN-BMD, 0.13 (0.01, 0.25) | ||
0.211 | LS-BMD, 0.082 (−0.045, 0.21) | ||||||||
2-h glucose | BMD | 6 SNPs | 2hGlu (133 010) BMD (32 961), adults; European | IVW approach | SD change in BMD per 1 mmol.L−1 increase in 2hGlu | 0.134 | FN-BMD, 0.089 (−0.027, 0.20) | ||
0.354 | LS-BMD, 0.06 (−0.06, 0.18) | ||||||||
Yang Q (2017)87 | Milk consumption | BMD | 1 SNP (lactose intolerance) | Milk consumption (32 961) BMD (32 961), adults; European | IVW approach | 1 SD change in BMD per 1 SD change of milk consumption | NA | FA-BMD, 0.049 (−0.128, 0.226) | No causal association between adult milk intake and BMD. |
NA | FN-BMD, −0.015 (−0.089, 0.059) | ||||||||
NA | LS-BMD, 0.015 (−0.073, 0.104) | ||||||||
Huang JV (2017)177 | Inflammation | BMD | 16 SNPs (CRP) | BMD (32 961); European | MR-Egger IVW approach | 1 g.cm−2 change in BMD per 1 log mg.L−1 change in total hsCRP | 0.506 | FA-BMD, 0.054 (NA) | No causal association between hsCRP and BMD |
0.726 | FN-BMD, −0.014 (NA) | ||||||||
0.184 | LS-BMD, −0.074 (NA) | ||||||||
Cousminer DL (2018)97 | Later puberty | BMD | 331 SNPs | 733 girls; European | Two-sample MR | 1 SD change in BMD per 1 year later onset of puberty | 0.004 6 | LS-BMD −0.179 | A causal association between later puberty and LS-BMD. |
Age at voice break | BMD | 43 SNPs | 685 boys; European | Two-sample MR | 1 SD change in BMD per 1 year earlier onset of age at voice break | 0.000 3; 7.04E-05 | LS-BMD −0.119; FN-BMD: −0.113 | A causal association between later puberty and LS/FN-BMD. | |
Larsson SC (2018)86 | Estradiol | BMD | 1 SNP | Estradiol (2 767) eBMD (32 965), adults; European | IVW approach | 1 SD change in BMD and g.cm−2 in eBMD per 10% increase in estradiol | 4.60E-06 | FN-BMD, 0.038 (NA) | A causal association between serum estradiol levels and increase BMD. |
0.001 | LS-BMD, 0.031 (NA) | ||||||||
6.00E-18 | eBMD, 0.030 (NA) | ||||||||
Vitamin D level | BMD | 5 SNPs | Vitamin D (42 274) BMD (32 961) eBMD (142 487), adults; European | IVW approach | 1 SD change in BMD per 1 SD change in 25OHD (g.cm−2 eBMD) | 0.37 | FN-BMD, 0.02 (−0.03, 0.07) | No causal association between vitamin D and BMD. | |
0.49 | LS-BMD, 0.02 (−0.04, 0.08) | ||||||||
0.02 | eBMD, -0.03 (−0.05, −0.01) | ||||||||
van Vliet NA (2018)99 | TSH | BMD | 20 SNPs | TSH (26 420) BMD (32 735), adults; European (mostly) | Two sample MR IVW approach | SD change in BMD per 1 SD decrease in serum TSH level | 0.92 | FN-BMD: 0.003 (−0.053, 0.048) | No causal association between serum TSH levels and BMD. |
0.73 | LS-BMD: 0.010 (−0.069, 0.049) | ||||||||
Guo R (2018)100 | Smoking status | BMD | 139-142 SNPs | Smoking status (32 735) BMD (445 921), adults; European (mostly) | Two-sample MR IVW approach | NA | 0.053 | FN-BMD: −0.139 (NA) | A causal association between smoking and decreased heel BMD. |
0.976 | LS-BMD: −0.003 (NA) | ||||||||
0.077 | FA-BMD: −0.264 (NA) | ||||||||
0.003 | Heel BMD: −0.053 (NA) | ||||||||
Alcohol consumption | BMD | 5–6 SNPs | Alcohol consumption (32 735) BMD (445 921), adults; European (mostly) | Two-sample MR IVW approach | NA | 0.964 | FN-BMD: −0.008 (NA) | No causal association between alcohol consumption and BMD. | |
0.742 | LS-BMD: 0.067 (NA) | ||||||||
0.593 | FA-BMD: 0.194 (NA) | ||||||||
0.822 | Heel BMD: 0.010 (NA) | ||||||||
Trajanoska K (2018)36 | T2D | Fracture | 38 SNPs | T2D: 56 862 (12 171 cases) Fracture: 185 057 cases, 377 201 controls; mostly European | Two sample MR; MR-Egger | OR of fracture per doubling in odds of T2D susceptibility | 0.37 | Fracture: 0.99 (0.99, 1.01) | No causal association between T2D and fracture risk. |
T1D | Fracture | 19 SNPs | T1D: 26 890 (9 934 cases) Fracture: 185 057 cases, 377 201 controls | Two sample MR; MR-Egger | OR of fracture per doubling in odds of T1D susceptibility | 0.57 | Fracture: 1.00 (1.00, 1.01) | No causal association between T1D and fracture risk. | |
Fasting glucose | Fracture | 35 SNPs | Fasting glucose (58 074) Fracture: 185 057 cases, 377 201 controls | Two sample MR; MR-Egger | OR of fracture per 1 SD increase in fasting glucose level | 0.24 | Fracture: 1.04 (0.97, 1.12) | No causal association of fasting glucose levels with fracture risk. | |
CAD | Fracture | 38 SNPs | CAD: 107 432 (41 513 cases) Fracture: 185 057 cases, 377 201 controls | Two sample MR; MR-Egger | OR of fracture per doubling in odds of CAD susceptibility | 0.76 | Fracture: 1.00 (0.99, 1.02) | No causal association between CAD and fracture risk. | |
Rheumatoid disease | Fracture | 30 SNPs | Rheumatoid disease: 58 284 (14 361 cases) Fracture: 185 057 cases, 377 201 controls; mostly European | Two sample MR; MR-Egger | OR of fracture per doubling in odds of rheumatoid disease susceptibility | 0.14 | Fracture: 1.01 (1.10, 1.02) | No causal association between rheumatoid disease and fracture risk. | |
Vitamin D | Fracture | 4 SNPs | Vitamin D: 33 996 Fracture: 185 057 cases, 377 201 controls; mostly European | Two sample MR MR-Egger | OR of fracture per 1 SD decrease in 25OHD | 0.07 | Fracture: 0.84 (0.70, 1.02) | No causal association of decreased 25OHD levels with increased fracture risk. | |
Dairy calcium intake | Fracture | 1 SNP (lactose intolerance) | Dairy calcium intake: 171 213 Fracture: 185 057 cases, 377 201 controls; mostly European | Two sample MR MR-Egger | OR of fracture per 1 SD increase in milk consumption | 0.94 | Fracture: 1.01 (0.80, 1.23) | No causal association between milk consumption and fracture risk. | |
FN-BMD | Fracture | 43 SNPs | FN-BMD: 32 961 Fracture: 185 057 cases, 377 201 controls; mostly European | Two sample MR MR-Egger | OR of fracture per 1 SD decrease in FN-BMD | <0.001 | Fracture: 1.55 (1.48, 1.63) | A causal association between decreased FN-BMD and increased fracture risk. | |
LS-BMD | Fracture | 40 SNPs | LS-BMD: 31 800 Fracture: 185 057 cases, 377 201 controls; mostly European | Two sample MR MR-Egger | OR of fracture per 1 SD decrease in LS-BMD | <0.001 | Fracture: 1.43 (1.37, 1.50) | A causal association between decreased LS-BMD and increased fracture risk. | |
Homocysteine | Fracture | 13 SNPs | Homocysteine: 44 147 Fracture: 185 057 cases, 377 201 controls; mostly European | Two sample MR MR-Egger | OR of fracture per 1 SD increase in homocysteine level | 0.78 | Fracture: 0.98 (0.92, 1.05) | No causal association between homocysteine level and fracture risk. | |
Inflammatory bowel disease (IBD) | Fracture | 151 SNPs | IBD: 34 652 (12 882 cases) Fracture: 185 057 cases, 377 201 controls; mostly European | Two sample MR MR-Egger | OR of fracture per doubling in odds of inflammatory bowel disease susceptibility | 0.92 | Fracture: 1.00 (1.10, 1.01) | No causal association between inflammatory bowel disease and fracture risk. | |
TSH | Fracture | 20 SNPs | TSH: 26 523 Fracture: 185 057 cases, 377 201 controls; mostly European | Two sample MR MR-Egger | OR of fracture per 1 SD decrease in serum TSH level | 0.78 | Fracture: 0.99 (0.94, 1.04) | No causal association between serum TSH levels and fracture risk. | |
Grip strength | Fracture | 15 SNPs | Grip strength: 142 035 Fracture: 185 057 cases, 377 201 controls; mostly European | Two sample MR MR-Egger | OR of fracture per 1 SD increase in grip strength | 0.01 | Fracture: 2.14 (1.13, 4.04) | A causal association between decreased grip strength and fracture risk. | |
Age of puberty | Fracture | 54 SNPs | Age of puberty: 182 416 Fracture: 185 057 cases, 377 201 controls; mostly European | Two sample MR MR-Egger | OR of fracture per 1 SD change, i.e., 3.9 years earlier menopause | 0.05 | Fracture: 1.10 (1.00, 1.21) | No causal association between earlier menopause and fracture risk. | |
Age at menopause | Fracture | 106 SNPs | Age at menopause: 69 360 Fracture: 185 057 cases; 377 201 controls; mostly European | Two sample MR MR-Egger | OR of fracture per 1 SD change, i.e., 1.4 years late puberty | 0.04 | Fracture: 1.06 (1.00, 1.13) | A causal association between late puberty and increased fracture risk. | |
Cerani A (2019)101 | Serum calcium | BMD | 1 SNP | Serum calcium: 61 079 BMD: 426 824; mostly European | IVW approach | 1 SD change in BMD per 1 SD change in serum calcium concentration | 0.85 | heel BMD, 0.003 (−0.059–0.066) | No causal association between serum calcium consumption and heel BMD. |
Serum calcium | Fracture | 1 SNP | Serum calcium: 61 079 Fracture: 76 549 cases, 470 164 controls; mostly European | IVW approach | OR of fracture per 1 SD increase in serum calcium concentration | 0.85 | Fracture, 1.01 (0.89–1.15) | No causal association between serum calcium consumption and fracture risk. | |
Xia (2020)88 | Psoriasis | eBMD | 60 SNPs | 301 667, European | One-sample MR | SD change in BMD per odds in log-odds of psoriasis susceptibility | 0.24 | heel BMD, − 0.04 (−0.11–0.029) | No causal association between psoriasis and heel BMD. |
Psoriasis: 19 032 cases, 286 769 controls eBMD: 462 824; European | Two sample MR | 0.28 | heel BMD, −0.002 (−0.009–0.002) | ||||||
Psoriasis | Fracture | 60 SNPs | Psoriasis: 19 032 cases, 286 769 controls Fracture: 45 087 cases, 317 775 controls; European | Two sample MR | OR of fracture per doubling in odds of psoriasis susceptibility | 0.72 | Fracture, 1.00 (0.99–1.02) | No causal association between psoriasis and fracture. | |
Psoriatic arthritis (PsA) | eBMD | 25 SNPs | 301 667; European | One-sample MR | SD change in BMD per odds in log-odds of psoriatic arthritis susceptibility | 0.88 | heel BMD, 0.002 (−0.025–0.030) | No causal association between psoriatic arthritis and heel BMD. | |
PsA: 3 061 cases, 13 670 controls eBMD: 462 824; European | Two sample MR | 0.69 | heel BMD, −0.001 (−0.005–0.003) | ||||||
Psoriatic arthritis | Fracture | 25 SNPs | PsA: 3 061 cases, 13 670 controls Fracture: 45 087 cases, 317 775 controls; European | Two sample MR | OR of fracture per doubling in odds of psoriatic arthritis susceptibility | 0.52 | Fracture, 0.99 (0.98–1.01) | No causal association between psoriatic arthritis and fracture. |
