Null Markers Research Articles

Logistic regression protects against population structure in genetic association studies

We conduct an extensive simulation study to compare the merits of several methods for using null (unlinked) markers to protect against false positives due to cryptic substructure in population-based genetic association studies. The more sophisticated "structured association" methods perform well but are computationally demanding and rely on estimating the correct number of subpopulations. The simple and fast "genomic control" approach can lose power in certain scenarios. We find that procedures based on logistic regression that are flexible, computationally fast, and easy to implement also provide good protection against the effects of cryptic substructure, even though they do not explicitly model the population structure.

Favorable treatment outcome of children with early stage large B-cell and anaplastic large cell lymphomas

8500 Background: The non-Hodgkin lymphomas (NHL) of childhood are heterogeneous. Large cell lymphomas (LCL) are relatively rare in children and sub-divided among diffuse large B-cell lymphomas (DLBCL), anaplastic large cell lymphomas (ALCL), peripheral T (PT) and other rare subtypes. Methods: We conducted a trial (POG 9219) for stage I and II NHL which accrued 396 children between 1992 and 1999. One hundred fifty-six (40%) had large cell lymphoma. All patients received nine weeks of chemotherapy including vincristine 1.5mg/m2 weekly for seven doses; doxorubicin 40mg/m2 and cyclophosphamide 750mg/m2 on days 1, 22 and 43; and prednisone 40mg/m2 daily for 28 days during the first 4 weeks and on days 43–47. None received radiotherapy. Results: Material for further study was available from 111 patients with LCL. Forty-six (42%) had DLBCL, 58 (52%) had ALCL, and 7 (6%) had PT. Among children with DLBCL, primary sites included lymph nodes (13), GI tract (8), bone (8), tonsil (8), testis (3), and other (6). Among children with ALCL, primary sites included lymph node (36), skin (11), bone (6), and other (5). Only one patient with DLBCL developed recurrent disease and died. The remaining 45 patients are all event free survivors. At 5 years, the projected event-free survival (EFS) is 98 % (SE 3%), and the overall survival (OS), 98 % (SE 3%). Thirty-five children with ALCL (60%) had T cell markers, and the remainder had null cell markers. Nine patients with ALCL (T=5; null=4) failed treatment: three failed induction, and six relapsed from complete remission, but were effectively salvaged. No patient with ALCL died. The projected 5 year EFS for early stage ALCL is 84 % (SE 7%) (DLBCL versus ALCL, p-value 0.02); the OS, 100%. Among seven patients with PT, two failed to achieve complete remission, one of whom has died. The remaining five are event-free survivors. Conclusions: Nine weeks of modest intensity chemotherapy are sufficient for children with early stage DLBCL. The outcome for ALCL and PT is less favorable than DLBCL, but most such children become long-term survivors. No significant financial relationships to disclose.

Case-control association studies with matching and genomic controlling.

Family-based association studies have gained in popularity for mapping disease-susceptibility gene(s) of complex diseases. However, recruiting family controls is often more difficult than recruiting unrelated controls. The author proposes a case-control study, where the possible biases due to population stratification are controlled by matching in the design stage and by genomic controlling in the data-analytic stage. The matching is based on a set of "stratum-delineating variables," such as, race, ethnicity, nationality, ancestry, and birthplace; and the genomic controlling is based on typing a number of null markers across the genome and applying the principle of multiplicative scaling of chi-square distribution. It pays to match carefully to have a higher proportion of correctly matched sets, as computer simulation showed that this would increase the power of the study. If matching is crude, one loses power but still has the correct type I error rate after genomic controlling. Power studies showed that the numbers of affected subjects required for the pair-matched study are comparable to those required by the case-parents design, if the study was conducted in a homogeneous population. As the (control-to-case) matching ratio increases, the number of affected subjects required decreases. With matching ratio tending toward infinity, the number required shrinks roughly by half. The case-control study with matching and genomic controlling frees us from family bondage, and the genetic problem as complicated as mapping genes can now be studied using simple epidemiologic methods.

Genetic association studies of adult-onset diseases using the case-spouse and case-offspring designs.

Genetic studies of complex human diseases rely heavily on the family-based association paradigm. However, recruiting parents or siblings can be a difficult task in practice. The author proposes two alternatives, the case-spouse and the case-offspring designs, that are to be analyzed by the mating disequilibrium test. Two assumptions are required: 1) the marker genotype frequencies at conception should be the same for both sexes; and 2) there is no selective attrition of marker allele(s) through gestation and over time. Within this setting, the case-spouse and the case-offspring studies are valid designs, even if only one sex can get the disease, even if cases/spouses/offspring all have different risk factor profiles, and even under assortative mating. If the population is stratified and there is intermarriage between strata, one can type additional null markers across the genome for an admixture correction. The number of families required in a case-spouse design is almost identical to that in a case-parents design. For the case-offspring study with one offspring per family, the number of families should be doubled. Because of the ease in recruiting control subjects, the case-spouse and the case-offspring designs are viable alternatives for genetic association studies of adult-onset diseases.