Table 4.

Propensity-score adjusted regression analysis results

Polyclonal immunoglobulin measures
Measure	Difference in serum levels [mg/dL] in relatives (95% CI)	p -Value
FLC(κ + λ)	− 0.2 (− 0.6,0.2)	0.33
IgA(κ + λ)	− 15.8 (− 41.6,10)	0.23
IgG(κ+λ)	− 9.8 (− 69.1,49.5)	0.75
IgM(κ+λ)	16.0 (1.3,30.6)	0.033
Monoclonal immunoglobulin measuresa
Measure	Fold difference for monoclonality in relatives (95% CI)	p -Value
FLC (κ/λ)	6.38 (1.65,34.97)	0.0063
IgA(κ/λ)	0.95 (0.3,2.76)	0.93
IgG(κ/λ)	1.44 (0.48,4.24)	0.51
IgM(κ/λ)	0.49 (0.12,1.52)	0.23

Propensity-score adjustment with matching on sex and age was used to control imbalances and nonlinearity in sex and age distributions between relatives and controls; p-values are from the Wald’s test. Polyclonal Ig variables are quantitative traits and are tested within a linear regression framework. Average differences in relatives compared with controls and 95% confidence intervals (CIs) are reported. Monoclonal Ig variables are dichotomous and tested within a logistic regression framework. Fold differences for monoclonality (abnormal κ/λ ratio, free, or bound) and 95% CI are reported from the logistic model. ^aAbnormality for FLC ratio (κ/λ) is based on its established diagnostic range (0.26–1.65). Abnormal HLC ratios are based on the 95% reference range from controls as follows: 0.84–1.91 for IgA(κ/λ); 1.02–2.87 for IgG(κ/λ); and 1.00–2.90 for IgM(κ/λ). Table entries that are bold and italic indicate those that are statistically different between at-risk relatives and controls