Identifying /Evaluating Sources Of Variability In Rodent

本项目的目标是确定和评价可能影响实验室研究中反应变异性的因素。例如，我们研究了新的NTP-2000饮食相对于旧的NIH-07饮食对F344大鼠体重、存活率和肿瘤发生率的影响。我们发现，与喂食NIH-07饮食的相应组相比，喂食NTP-2000饮食的组的存活率显著（p <0.05）更高，与性别或饲养条件无关。使用NTP-2000饮食还与两种性别中垂体瘤的发生率降低以及雄性中肾上腺嗜铬细胞瘤和包皮腺肿瘤的发生率降低相关。在接受NTP-2000饲料的动物中，肾病的发生率和严重程度也降低，尤其是雄性。与给予NIH-07饲料的大鼠相比，肾病严重程度降低和脑垂体肿瘤发生率降低可能是导致接受NTP-2000饲料的大鼠存活率提高的主要因素。本研究的次要目的是比较饲喂研究对照组（分组饲养在悬挂抽屉式聚碳酸酯笼中，接受糊状饲料）和吸入研究对照组（单独饲养在不锈钢丝网笼中，饲喂颗粒饲料）的肿瘤发生率。这两个对照组在各种肿瘤的发生率方面存在显著差异（p <0.01），表明动物护理和饲养方案的差异可能影响F344大鼠中的肿瘤发生。我们还评估了对数线性外推模型，该模型已在一系列文章中描述为可以证明阈值并解决与低剂量癌症风险外推相关的不确定性的程序。我们指出了这一模型的重大缺陷，即它（i）是一个基本上强制阈值的模型;（ii）假设控制响应是已知的，没有错误;（iii）忽略了外推阈值估计值的不确定性;（iv）主观地丢弃外推的数据点，包括低剂量下可能重要的肿瘤响应;（五）据称即使在所有剂量组都显示出高度显著的致癌作用时，也能明确地证明阈值;（vii）在低剂量区域的某个未知点失效，可能远高于声称的阈值剂量;（viii）可证明显著低估低剂量风险。因此，科学界采用这种对数线性外推模型进行化学致癌风险评估将是一个严重的错误。

The goal of this project is to identify and evaluate factors that may influence variability in response in laboratory studies. For example, we investigated the impact of the new NTP-2000 diet on body weight, survival and tumor occurrence in F344 rats relative to the older NIH-07 diet. We found that survival was significantly (p<0.05) higher in groups fed the NTP-2000 diet compared to the corresponding groups fed the NIH-07 diet, irrespective of sex or housing conditions. Use of the NTP-2000 diet was also associated with a decreased incidence of pituitary gland tumors in both sexes and decreased incidences of adrenal pheochromocytoma and preputial gland tumors in males. The incidence and severity of nephropathy was also reduced in animals receiving the NTP-2000 diet, especially males. The decreased nephropathy severity and the decreased incidence of pituitary gland tumors are likely the major factors contributing to the improved survival of rats receiving the NTP-2000 diet compared to those given the NIH-07 diet. A secondary objective of this study was to compare tumor incidences in feeding study controls (which are group housed in hanging-drawer-type polycarbonate cages and receive the diet in mash form) and inhalation study controls (which are individually housed in stainless steel wire mesh cages and are fed pelleted feed). These two control groups differed significantly (p<0.01) in the incidence of a variety of tumors, suggesting that differences in animal care and housing protocols may impact tumor occurrence in F344 rats. We also evaluated a log linear extrapolation model that has been described in a series of articles as a procedure that can demonstrate a threshold and resolve the uncertainies associated with low dose cancer risk extrapolation. We pointed out the significant shortcomings of this model, namely that it (i) is a model that essentially forces a threshold; (ii) assumes the control response is known without error; (iii) ignores the uncertainty of the extrapolated threshold estimate; (iv) subjectively discards datapoints from the extrapolation, including potentially important tumor responses at low doses; (v) purportedly can unequivocally demonstrate thresholds even when all dosed groups show highly significant carcinogenic effects; (vi) purportedly can unequivocally demonstrate thresholds with as few as two doses; (vii) fails at some unknown point in the low dose region, possibly well above the purported threshold dose; and (viii) demonstrably can significantly under-estimate low dose risk. For these reasons, it would be a serious mistake for the scientific community to adopt this log linear extrapolation model for chemical carcinogenesis risk assessment.