To determine the safety of one intravenous injections of autologous TGFBeta-resistant CTLs directed to EBV through their native receptor and HER2 through their CAR in patients with advanced HER2-positive lung tumors, the investigators implement a Bayesian dose-finding phase-I trial design, called the modified continual reassessment method (mCRM). Each patient will receive one injection of the TGFBeta resistant HER2/EBV-specific CTLs according to the dosing schedule (nine levels) identified in Section F2 of the Protocol Summary.

Each patient will be followed for 6 wks after the CTL infusion for evaluation of dose limiting toxicity (DLT). For this trial, the maximum tolerated dose (MTD) is defined to be the dose which causes DLT in any patients. The toxicity will be evaluated by the NCI CTCAE Version 3.0. Any Grade 3, 4 or 5 toxicity primarily related to the CTL infusion will be defined as a DLT. Any such toxicities where there is a question of causality will be discussed with the FDA. All patients within a dose cohort should have completed the 6-week window after the CTL infusion for assessment of dose limiting toxicity as defined below prior to enrollment of patients into the next recommended dose level.

The modified continual reassessment method will be implemented with a cohort of size 2 based on the exponential dose-toxicity model. The fixed prior probabilities of DLT for the nine dose levels indicated above were assumed to be 0.1%, 0.2%, 1%, 2%, 5%, 8%, 10%, 15%, and 21% respectively. To ensure patient safety, the mCRM starts from the lowest dose level and limits the dose escalation to one dose level at a time. DLT event in the lower dose cohort will be used to update the dose-toxicity curve. The next patient cohort is assigned to the dose level with an associated probability of DLT closest to the target probability of 20%. This process continues until at least 22 patients have been accrued into the trial or 6 patients have been treated at the current MTD. Depending on patient availability and dose escalation, a maximum of 40 patients will be recruited into this Phase I trial. The final MTD will be the dose with probability closest to the target toxicity rate at these termination points. Since TGFBeta resistant HER2/EBV-CTL generation for this trial takes 4 to 5 months the investigators plan to generate CTL products for up to 67 patients accounting for potential 'drop outs' and change in patients' eligibility status (60% completion rate).

The investigators did simulations with 10,000 replications in order to compare the operating characteristics of mCRM with a standard 3+3 dose-escalation design. Compared to the 3+3 design, the proposed mCRM design allows a better estimate of the MTD based on a higher probability of declaring the third dose level as the MTD, allocated a smaller number of patients at lower and likely ineffective dose levels, afforded a slightly lower average total number of patients needed to complete the trial (11 vs 12), and yielded similar toxicities on average compared to the standard 3+3 design (1.5 for both). The investigators expect a shallow dose-toxicity curve and feel comfortable with slightly more accelerated dose-escalations without a substantial compromise in patient safety.

During the study, real-time monitoring of patient toxicity outcome will be performed in order to implement estimation of the dose-toxicity curve and determine the dose level for the next patient cohort using one of the pre-specified dose levels. Patients in a cohort can be enrolled concurrently but both patients within a dose cohort should have completed the 6-week window after the CTL infusion for assessment of DLT prior to enrollment of patients into the next recommended dose level.

A separate criteria for halting the trial will be employed based on incidence of lymphoproliferation. Specifically, accrual will be stopped if one patient experiences lymphoproliferation at any time point during the study.