Imputing physical health status scores missing owing to mortality: results of a simulation comparing multiple techniques.

BACKGROUND: Having missing data complicates the statistical analysis of health-related quality-of-life (HRQOL) data and, depending on the extent and nature of missing data, can introduce significant bias in treatment comparisons. OBJECTIVE: We evaluated the bias associated with 4 different imputation methods for estimating physical health status (PHS) scores missing as a result of mortality. METHODS: A simulation study was conducted in which we systematically varied mortality rates from 0% to 30% and change in PHS scores from -20 to 20 on a 100-point scale for a 2-group clinical trial with follow-up over 18 months. The 4 imputation methods were last value carried forward (LVCF), arbitrary substitution (ARBSUB), empirical Bayes (BAYES), and within-subject modeling (WSMOD). Pseudo-root mean square residuals (RMSRs) and differences between true and estimated slopes were used to evaluate how well the imputation methods reproduced the true characteristics of the simulated population data. RESULTS: ARBSUB and BAYES methods have the smallest RMSRs compared with LVCF and WSMOD across all mortality rates. As the rate of missing data resulting from mortality increased, all imputation techniques deviated more from population data. The BAYES technique was best at reproducing group slopes in cases with differential mortality rates or when mortality rates exceeded 15%. WSMOD and LVCF significantly underestimated changes in PHS. CONCLUSIONS: The different imputation methods produced comparable results when there were few missing data. The BAYES approach most closely estimated true population differences and change in PHS regardless of missing data rates. These findings are limited to physical health and functioning measures.

Eicosanoids and linoleate-enhanced growth of mouse mammary tumor cells.

Metastatic mouse mammary tumor cell line 4526 was used to determine whether linoleate (LN)-derived cyclooxygenase metabolites were involved in the mechanism of LN-enhanced 4526 tumor growth. Unstimulated line 4526 cells converted LN to both PGE1 and PGE2 in serum free medium (SFM). However, neither prostaglandin (PG) influenced growth, while db-cGMP, but not db-cAMP, stimulated growth to the same extent as LN. Cyclooxygenase inhibitors stimulated growth while suppressing PG synthesis. Lipoxygenase inhibitors decreased growth in a dose dependent manner. Supplemental LN had no effect on cyclooxygenase inhibition while the IC50s for lipoxygenase inhibition were increased several fold. These results indicate that lipoxygenase products rather than cyclooxygenase metabolites play a major role in LN-stimulated growth of line 4526 cells.

Relationship of the uptake and beta-oxidation of 18-carbon fatty acids with stimulation of murine mammary tumor cell growth.

The effects of stearic (18:0), oleic (18:1) and linoleic (18:2) acid on the in vitro growth of murine mammary tumor cell line 4526 were compared to 4526 uptake and beta-oxidation of those fatty acids. Both 18:1 and 18:2 stimulated, while 18:0 inhibited growth. Likewise, uptake and oxidation rates were much lower for 18:0 than for 18:1 and 18:2. The influence of glucose, insulin and fatty acid-depletion on 4526 fatty acid incorporation and oxidation was also determined. The correlation between effects on in vitro growth and ability to assimilate 18:0, 18:1 and 18:2 may partially explain the stimulation by unsaturated and inhibition by saturated fatty acids of in vivo 4526 tumorigenesis.

Platelet activating factor as a proinflammatory mediator in acetic-induced colitis in the rat.

Platelet activating factor (PAF) induces neutrophilia and produces a variety of gastrointestinal lesions. The role of PAF as a proinflammatory mediator was examined by measuring the production of PAF and the efficacy of the PAF antagonists WEB 2086 and Ro 24-0238 in acetic acid (HOAc)-induced colitis. PAF levels within the colonic mucosa, as measured by radioimmunoassay, increased from 259 +/- 119 ng/mg in control tissue, to 616 +/- 266 ng/mg in HOAc inflamed tissue. The accumulation of neutrophils within the mucosa was decreased by 53 +/- 10% by pretreatment with 3 mg/kg WEB 2086, and by 43 +/- 11% by 3 mg/kg Ro 24-0238. PAF-induced neutrophilia had no effect on the severity of HOAc-induced colitis, therefore, PAF my be involved in sustaining the chronic inflammation of colitis.

Modulation of mouse mammary tumor growth and linoleate enhanced metastasis by oleate.

This study examines whether oleate may influence the linoleate enhanced metastasis of line 4526 murine mammary tumors. In addition, the in vitro proliferative response of line 4526 to oleate and other selected fatty acids was assessed. Initially, the tumor cells were grown in a defined medium supplemented with palmitate, stearate, oleate, linoleate, linolenate or arachidonate. The unsaturated fatty acids stimulated and the saturated fatty acids inhibited proliferation compared to fatty acid-free medium. Next, we examined the effect of oleate on the linoleate enhanced metastasis of 4526 tumors by substituting oleate for saturated fat in isoenergetic diets containing high or low levels of linoleate. Oleate had no effect on metastasis in mice fed the high linoleate diets but it significantly increased metastasis in mice fed the low linoleate diets. Finally, the fatty acid compositions of tumors and mammary fat pads were compared to diet fatty acid compositions and metastatic frequency. Metastasis corresponded more closely to total unsaturated fatty acids than to total polyunsaturated fatty acids or to any individual fatty acid. These studies suggest that both mono- and polyunsaturated fatty acids may stimulate mammary tumor metastasis. However, the influence of dietary oleate probably depends on the level of linoleate and total unsaturated fatty acids in the diet.

Linoleic acid metabolism in metastatic and nonmetastatic murine mammary tumor cells.

The mechanism(s) by which dietary linoleic acid (18:2n-6) enhances mammary tumor growth and metastasis is not known. Since arachidonic acid (20:4n-6)-derived prostaglandins (PG) may play a role in the metastatic dissemination of tumor cells, the ability of two murine mammary tumor cell lines, 4526 (metastasis positive) and line 168 (spontaneous metastasis negative), to convert 18:2n-6 into prostaglandins was examined. Cells were initially incubated with [14C]18:2n-6 and after 8-24 h the [14C]fatty acids were quantitated by high-performance liquid chromatography following transesterification. [14C]18:2n-6 was metabolized primarily to [14C]dihomogammalinolenic acid (20:3n-6) in line 4526 cells and [14C]20:4n-6 in line 168 cells. Examination of cellular fatty acid levels revealed a 20:3n-6/20:4n-6 ratio of 1.79 +/- 0.36 and 0.20 +/- 0.02 in line 4526 and 168 cells, respectively. These data are consistent with an inherently lower delta 5 desaturase activity in line 4526 relative to 168. To assess the metabolism of 18:2n-6 into eicosanoid products, the cell lines were prelabeled with [14C]18:2n-6 or 0-40 microM nonradiolabeled 18:2n-6 overnight and subsequently stimulated with calcium ionophore A23187 for 1 h. Total PGE production, as determined by radioimmunoassay, was greater in 168 relative to 4526 cells at all 18:2n-6 concentrations. 14C-prostaglandins detected by high-performance liquid chromatography and argentation thin-layer chromatography were: PGF1 alpha and PGE1 (derived from 20:3n-6) and PGF2 alpha and PGE 2 (derived from 20:4n-6) from line 4526; PGE1 and PGE2 from line 168. PGE1/PGE2 ratios were 1.43 +/- 0.07 and 0.23 +/- 0.03 for 4526 and 168 lines, respectively. Neither cell line synthesized lipoxygenase products following [14C]18:2n-6 or [3H]-20:4n-6 incubations under the conditions employed. Additional studies are warranted in order to define the biological properties of 1- and 2-series cyclooxygenase products as they relate to tumor cell metastasis.

Dietary fat modulation of murine mammary tumor metabolism studied by in vivo 31P-nuclear magnetic resonance spectroscopy.

The effect of dietary fat concentration and saturation on high energy phosphate metabolites and phospholipid turnover in transplanted line 168 murine mammary tumors was studied using surface coil 31P-nuclear magnetic resonance spectroscopy. Female BALB/c mice were fed one of five diets each containing at least the minimum of essential fatty acids (EFA). Four diets contained additional safflower or palm oil for a total fat concentration of 5 or 20% by weight. The growth rate of tumors from mice fed the high safflower oil diet was significantly greater than the growth rate of tumors for mice fed all other diets including the one which contained the minimal EFA. 31P-nuclear magnetic resonance-observable phosphate metabolite ratios. ATP/Pi, ATP/phosphomonoester (ATP/PME), and PME/Pi, and tumor pH of line 168 tumors decreased with increasing tumor volume, indicating a shift from active to inactive tumor metabolism. The rates of those decreases with progressive tumor growth differed significantly among tumors of mice fed the different diets. Decreases in ATP/Pi, ATP/PME, and pH were the most rapid in the tumors of mice fed the high safflower oil diet and significantly faster than tumors of mice fed the diet containing minimum EFA. In addition, the decrease in the PME/Pi ratio of tumors was significantly greater in mice fed the high fat (high palm oil and high safflower oil) diets than mice fed the diet containing the minimum of EFA. The rate of decline of ATP/Pi and ATP/PME with progressive tumor growth was directly correlated with levels of linoleic acid as well as total unsaturated fat. High levels of a polyunsaturated fat had a significant effect on mammary tumor metabolism particularly during early stages of tumor growth. Differences in high energy phosphate metabolite dynamics relative to dietary fat were present in tumors of equal volume. Thus, dietary fat influences on mammary tumorigenesis may be related to high energy phosphate metabolites.