Assessing a PET-based technique for characterizing beta cell mass in vivo

这份年度报告中描述的工作都源于哥伦比亚大学的Paul Harris博士的观察，他使用差异基因表达研究指出，β细胞(比其他胰腺细胞在更大程度上)表达囊泡单胺转运体-2(VMAT-2)。Harris博士和他的同事继续使用名为11C-二氢四苯并嗪(11C-DTBZ)的VMAT-2放射性标记配体作为显像剂，在T1 DM大鼠模型上进行正电子发射断层扫描(PET)研究，结果表明，随着β细胞质量在自身免疫过程中下降，11C-DTBZ-PET胰腺信号也随之下降。事实上，在该模型中，PET信号似乎能够预测糖尿病的发生。为了开发这项技术应用于临床介入试验，招募有T1糖尿病风险的人类，我们启动了几项临床前研究，这些研究在工作单位编号DK055114-01的年度报告中描述。我们临床前研究的早期令人鼓舞的结果，加上11C-DTBZ-PETS在密歇根大学进行的跟踪帕金森病患者多巴胺能神经细胞团的临床研究中证明的安全性记录，鼓励我们启动研究，将11C-DTBZ-PET在三个研究组中产生胰腺信号：(1)正常对照组，(2)长期存在T1 DM且几乎没有内源性胰岛素产生能力的患者，(3)多年前成功进行胰腺移植治疗的T1 DM病史的患者。 在机构审查委员会适当批准后，我们首先评估了5名血糖控制正常的受试者，这5名受试者的口服葡萄糖耐量试验(OGTT)正常。所有5名受试者的11C-DTBZ-PET胰腺信号都很强，个体间差异很小。接下来，我们获得了5名有成功接受胰腺移植治疗的T1 DM病史的受试者的11C-DTBZ-PET图像(这些受试者也在工作单元编号DK055111-01的年度报告中描述)，原因是这些人有两个胰腺处于相同的代谢条件下，即他们的原始胰腺分泌非常少的胰岛素，因此预计包含很少的β细胞，以及移植的胰腺具有足够的β细胞功能，以将患者的血糖维持在正常范围内(同样通过OGTT)。这5名受试者的11C-DTBZ-PET扫描显示，自体胰腺上的信号强度很低或没有，而移植胰腺上的信号要强得多(尽管信号强度略低于非糖尿病受试者)。第三组由长期患有T1糖尿病的受试者组成。这组受试者很少或没有C-肽反映他们预期的严重减少的β细胞质量，但与预测相反，12名受试者中的每个人都显示出相当强的11C-DTBZ-PET产生的胰腺信号。事实上，在两名患有T1 DM的受试者中，11C-DTBZ-PET产生的胰腺信号与我们的非糖尿病对照组观察到的信号一样大，甚至更高。 为了评估糖尿病受试者的胰腺PET信号是否会被异常高的非特异性11C-DTBZ结合所混淆，我们利用了这样一个事实，即只有阳性的11C-DTBZ对映体与VMAT2结合。我们在一名血糖正常的受试者和一名长期患有T1 DM的受试者中重复了DTBZ-PET扫描。在这两种情况下，我们将产生的PET信号与阳性和阴性的11C-DTBZ对映体进行比较。如果在T1 DM患者中观察到的强胰腺信号是非特异性DTBZ摄取的继发性信号，我们预计在T1 DM患者中将观察到类似的信号，而不考虑DTBZ对映体，而对于正常对照组，我们预计DTBZ对映体阳性将产生明亮的胰腺信号，而阴性对映体将产生更弱的信号。令我们惊讶的是，无论是正常对照组还是长期患有T1 DM的受试者，11C-DTBZ阴性对映体产生的胰腺PET信号至少与11C-DTBZ阳性对映体产生的信号一样明亮。我们的结论是，尽管在临床前研究中显示了希望，11C-DTBZ-PET产生的胰腺信号与BCM没有很好的相关性。我们正在准备一份手稿来报告这些令人失望的观察结果。

The work described in this annual report all emanated from observations made by Dr. Paul Harris at Columbia University who, using differential gene expression studies, noted that beta cells (to a much greater degree than other pancreatic cells) express the vesicular monoamine transporter-2 (VMAT-2). Dr. Harris and colleagues went on to perform studies using a VMAT-2 radiolabeled ligand named 11C-dihydrotetrabenazine (11C-DTBZ) as an imaging agent for positron-emission tomography (PET) studies in a rat model for T1DM with results suggesting that as beta cell mass fell during the autoimmune process, so did the 11C-DTBZ-PET pancreatic signal. Indeed, in that model the PET signal seemed able to predict diabetes onset. With an eye toward developing the technique for application in clinical interventional trials enrolling humans at risk for T1DM, we initiated several pre-clinical studies described in the annual report for work unit number DK055114-01. Early promising results from our pre-clinical studies, coupled with 11C-DTBZ-PETs proven safety record in clinical studies performed at the University of Michigan to follow dopaminergic neuronal cell mass in subjects with Parkinsons disease, encouraged us to initiate studies correlating 11C-DTBZ-PET generated pancreatic signals in three study groups: (1) normal control subjects, (2) subjects with long-standing T1DM and little to no endogenous insulin producing capacity and (3) subjects with a history of T1DM successfully treated years earlier with a pancreas transplant. After appropriate Institutional Review Board approval, we first evaluated 5 subjects with normal glycemia control as evidenced by normal oral glucose tolerance tests (OGTT). The 11C-DTBZ-PET generated pancreatic signal from all 5 subjects was strong with little inter-individual variability. We next obtained 11C-DTBZ-PET images from 5 subjects with a T1DM history successfully treated with a pancreas transplant (subjects also described in the annual report for work unit number DK055111-01) reasoning that such individuals have two pancreases under identical metabolic conditions, i.e. their original pancreas secreting very little insulin and therefore expected to contain few beta cells, and the transplanted pancreas with sufficient beta cell function to maintain the patients' blood glucose in the normal range (again by OGTT). The 11C-DTBZ-PET scans from these 5 subjects revealed little to no signal intensity over the native pancreas, with much stronger signals over the pancreas allograft (albeit slightly less intense than the non-diabetic subjects pancreatic signals). The third group consisted of subjects with long-standing T1DM. The subjects in this group made little to no C-peptide reflecting their expected severely reduced beta cell mass, but contrary to prediction, each of the 12 subjects displayed quite strong 11C-DTBZ-PET generated pancreatic signals. In fact, in two subjects with T1DM, the 11C-DTBZ-PET generated pancreatic signal was as great as or even greater than the signal observed in our non-diabetic control subjects. In order to evaluate whether the diabetic subjects pancreatic PET signals might be confounded by unusually high non-specific 11C-DTBZ binding, we took advantage of the fact that only the positive 11C-DTBZ enantiomer binds to VMAT2. We repeated DTBZ-PET scans in one subject with normal glycemia, and in one subject with long-standing T1DM. In both cases, we compared PET signals generated with the positive and negative 11C-DTBZ enantiomers. If the strong pancreatic signals observed in the subjects with T1DM were secondary to non-specific DTBZ uptake, we expected to observe comparable signals in the subject with T1DM regardless the DTBZ enantiomer, while for the normal control subject, we expected the positive DTBZ enantiomer to generate a bright pancreatic signal, and the negative enantiomer to generate a much weaker signal. To our surprise, for both the normal control subject and the subject with long-standing T1DM, the pancreatic PET signal generated by the negative 11C-DTBZ enantiomer was at least as bright as the signal generated by the positive 11C-DTBZ enantiomer. We conclude that despite the promise displayed in pre-clinical studies, 11C-DTBZ-PET generated pancreatic signals do not correlate well with BCM. We are preparing a manuscript to report these disappointing observations.