Soon after the methodology for engrafting a chimeric antigen receptor with antibody-like specificity into T cells became available, Rosenberg’s group at the NCI in collaboration with Eshhar at the Weizmann Institute demonstrated anti-tumor activities of adoptively transferred T cells made to express chimeric antibody/T cell receptors targeting folate receptors in a nude mouse model [
12]. Thereafter, Rosenberg’s group translated the idea into a phase I clinical trial of CAR-T cells targeting folate receptors in ovarian cancer [
13]. This trial did not lead to anti-tumor effectiveness, but showed that CAR-T cells can be relatively safely administered to human patients. Thereafter, in another early clinical trial of CAR-T cells engineered to express specificity for the cell adhesion molecule, L1 (CD171), in neuroblastoma, Park et al. [
14] also could not demonstrate objective clinical responses, but observed that the adoptively transferred cells exhibited poor persistence. Shortly thereafter, Till et al. [
15] from the City of Hope and the Fred Hutchinson Cancer Research Center also reported a proof-of-concept for tumor specific chimeric T cell receptor-based strategy in the treatment of non-Hodgkin’s lymphomas and Pule et al. [
16] reported anti-tumor activities in neuroblastoma with CAR-T cell therapy targeting GD2 as a tumor-associated antigen.
Interest in tumor immunotherapy with CAR-T cells thereafter shifted to treating B cell malignancies targeting CD20/CD19. Using CD20/CD19 specific CAR-T cells, Jensen et al. [
17] also found poor persistence of the CAR-T cells in patients due to anti-transgene rejection. Almost simultaneously, Kochenderfer et al. [
18]; however, reported complete remission of CD19 positive advanced follicular lymphoma with the infusions of autologous CD19 CAR-T cells administered with IL-2. Interestingly, the treatment with CD19 CAR-T cells also led to the depletion of the patient’s B cell lineage from blood and bone marrow [
18]. Soon, several groups confirmed these observations and reported that CAR-T cell therapy targeting CD19 could achieve complete remissions, including molecular remissions, in a fair fraction of CD19 positive leukemia that were refractory to traditional systemic treatments [
19‐
23]. Most importantly, multicenter trials confirmed these basic observations and long-term follow-ups revealed that the complete remissions could be durable [
24,
25] and raised the prospect of achieving “cures” of a sizeable fraction of refractory lymphocytic leukemia patients with CD19 CAR-T cell-based therapy [
5,
26]. While this was clearly a positive development, CD19 CAR-T cell therapies came with substantial toxicities including fatalities (discussed later).
Design of CAR-T cells, dose schedule, additional measures, et cetera
A number of issues on CAR-T cell-based therapies remain unsettled. Among others, these include: (1) the ideal design of CAR-T cells (the inclusion of one or more than one co-stimulatory molecules and other molecules potentially affecting T cell function, survival, etc., viral vs. non-viral mechanism for gene transfer into T cells, ex vivo expansion methodology, et cetera); (2) the number of cells to be infused and the number of infusions for maximum benefit; (3) the need for and the nature of conditioning regimen (4), cytokine support to facilitate persistence of the infused cells, et cetera. While these issues are yet to be settled, most clinical trials have used the second generation CAR-T cells utilizing viral as well as non-viral gene transfer techniques, some form of pre-therapy lymphodepletion measure, exogenous IL-2 support or no cytokine support, and varied number of cells (10
5–10
9 cells per infusion) [
5,
26]. Of note, CAR-T cell-based trials at multiple centers have tested infusions of a wide range of cell numbers and objective responses, including complete remissions, have been observed with different cell numbers [
5,
26]. It is; however, by no means clear that large numbers of cells are needed to obtain complete remissions [
5,
26]. In fact, complete remission has been achieved in chronic lymphocytic leukemia with as little as 1.4 × 10
5 antigen specific population split into three consecutive days [
19].
Adverse events with CAR-T cell therapy and risk–benefit analysis
The side effects of CAR-T cell therapies can vary from mild constitutional syndrome (Flu-like symptoms, malaise, fever, etc.) to life-threatening situations requiring intensive care [
5,
26‐
30]. The major toxicities usually result from: (a) inflammatory cytokines elaborated by the infused T cells from their cognate interactions with the target epitopes, (b) from secondary immune activation and from the activation of macrophages (usually referred to as “cytokine release syndrome or CRS” and macrophage activation syndrome), and (c) from massive target cell lysis (usually referred to as “tumor lysis syndrome”) resulting in high fever, hypotension, hypoxia, intravascular coagulation, etc. [
5,
26‐
30]. The side effects can result from the CAR-T cells acting against epitopes “on target” as well as “off target”. At times, the toxicities can occur quickly, can lead to various organ (hepatic, gastrointestinal, respiratory, cardiovascular, endocrine, and neurological) dysfunctions, and at times they can be devastating resulting in death. Indeed, fatal pulmonary complication (such as seen with CAR-T targeting ERBB2 [
31]), and a number of deaths from neurological complications—evidently from cerebral edema—lately encountered in the ongoing Juno Therapeutics trial. Although the reason for the fatal neurological side effects is yet to be fully clarified, the fatalities were attributed to Fludarabine used as a part of the preparative regimen.
Understandably, CAR-T cell therapy requires careful monitoring for side effects; and depending on the nature of the side effects, adverse reactions from CAR-T cell therapies call for quick supportive cares including intensive cares, as well as additional measures when needed. Additional measures for the major toxicities include the use of systemic steroids and the anti-IL-6 receptor blocking monoclonal antibody, tocilizumab, for CRS-related toxicities [
25‐
28].
It should be pointed out that while the various serious adverse events following CAR-T cell therapy remain a matter of concern, most of the toxicities from CD19 CAR-T cell therapy are becoming “manageable” [
5,
8,
26‐
30]. However, as CAR-T cell therapies designed to target other antigens in other tumor types enter into the clinics, they are likely to lead to different types of “on target and off target” adverse events. Further, as different pre-conditioning regimens get introduced, they can also lead to additional toxicities and can add to the side effects from the inflammatory responses by the infused T cells. Thus’, CAR-T cell therapy essentially navigates on uncharted territories. As such, a true evaluation of risk–benefit analysis of CAR-T cell therapy is presently not possible. Nonetheless, the benefits of durable complete remissions that a fair fraction of patients with B cell malignancies who have exhausted all standards of care can now enjoy with this form of therapy represent a remarkable development in cancer immunotherapy.
The development of ICI-based cancer immunotherapies
In the early days of studies of signaling pathways in T cells, CTLA-4 was thought to be just another co-stimulatory receptor [
32]. It was; however, soon learned that the ligation of CTLA-4 with an agonist, in fact, opposes CD28-driven co-stimulation and results in the down-regulation of T cell responses [
33,
34]. As the role of CTLA-4 functioning as a brake in T cell responses became better understood, Allison recognized that taking the brake off from T cells might help unleash anti-tumor effector responses. Pursuing the idea in animal models, his group showed that the antibody mediated blockade of CTLA-4 indeed leads to tumor regression [
35]. It took some time to take the idea to the bedside, but eventually the anti-CTLA-4 antibody was humanized, industries got involved, and humanized anti-CTLA-4 antibody was moved to clinical trials.
Several phase I–II trials quickly revealed anti-tumor activities of the inhibitory anti-CTLA-4 antibody [
36‐
39] opening a novel strategy in cancer immunotherapy [
40]. Interestingly, while several types of tumors showed positive response, the results of CTLA-4 blockade were most impressive in melanoma. Trials of anti-CTLA-4 blockade-based therapy continued and several multicenter-based trials of Ipilimumab, alone or in combination with gp100-based immunization, confirmed anti-tumor activities of the anti-CTLA-4 antibody Ipilimumab alone, and most importantly, showed its effectiveness in improving survival of melanoma patients [
41].
Yet interestingly, the best overall response rates (95% CI) from the multicenter trials were not all that high (The response rates for Ipilimumab, gp100, and Ipilimumab + gp100 were 10.9, 1.5, and 5.7%, respectively). No complete response was noted with gp100 alone, and the CR rate for Ipilimumab and Ipilimumab + gp100 were 2 and 1%, respectively. However, the median survival rates were improved with Ipilimumab alone, gp100 alone, or with Ipilimumab + gp100 (The median survivals of the three groups were 10.1, 6.4 and 10 months, respectively). More importantly, the overall survivals of the three groups at 2 years (Kaplan–Meier survival curves) for the Ipilimumab, gp100, and Ipilimumab + gp100 groups were 23.5, 13.7 and 21.6%, respectively [
41]. Further, the disease was stabilized in a significantly larger fraction of patients receiving Ipilimumab alone or receiving Ipilimumab + gp100 and objective responses in a fraction of patients receiving re-induction therapy either with Ipilimumab alone or with Ipilimumab + gp100, were noted. The response rates of the NCI trials were not substantially different. This trial showed that Ipilimumab alone can induce complete responses in a fraction of patients and that the combinations of Ipilimumab with IL-2 or with gp100 might be little bit better [
42]. Understandably, the data from the multicenter trials showing improvement in response rate and improvement in overall survival with Ipilimumab [
41] generated considerable excitement in the field.
Subsequently, additional trials were launched and outcome analyses of the initial trials continued. The analyses of ten subsequent prospective trials and two retrospective studies involving little over 1800 patients revealed overall survival rates of 24 and 26% at 3 years receiving Ipilimumab + gp100 or Ipilimumab alone with a fraction of patients living at the 10 years’ mark [
43]. Similarly, a long-term follow-up of the NCI trial also revealed “durable” and “potentially curative tumor regressions” in melanoma with the combinations of Ipilimumab and IL-2 or Ipilimumab and gp100 (5-year survival rates of 25 and 23%, respectively, while the same with Ipilimumab alone being 13%) [
43]. Of considerable interest, some patients in the NCI trial had shown continued tumor regressions over time eventually achieving complete tumor regressions even after stoppage of the treatment. This follow-up also showed that although relapses in some patients were noted after 3–4 years, the survival curve flattened at 3–4 years’ mark [
43].
The development of ICI targeting PD-l and PD-L1 followed a similar path. Initially in a search for the genes involved in programmed cell death, Honjo’s group identified PD-1 as a programmed death associated molecule, hence the name programmed death for PD [
44]. Thereafter Sharpe and Freeman, in collaboration with Honjo, discovered the natural ligands for PD-1—i.e., PD-L1 and PD-L2—and found that the engagement of PD-1 by the ligands negatively regulate T cell function pretty much the same way as CTLA-4 but, in a non-overlapping manner [
45]. Around the same time, Chen’s group independently discovered PD-L1 that they called B7-H1 as a third co-stimulatory member of the B-7 family [
46]. They went on to show that tumors cells often express the B7-H1/PD-1 ligand seemingly to escape host immune responses and recognized the potential for designing T cell-based cancer immunotherapy [
47]. These collective observations led to the developments of reagents that could target PD-1 as well as PD-L1 as another approach to inhibitory receptor blockade-based tumor immunotherapy. In due time, phase I trials began with the anti-PD-1 antibody Nivolumab and the anti-PD-L1 antibody BMS-936559. Both trials included several different types of tumors and anti-tumor activities with both reagents were shortly reported [
48,
49]. Another anti-PD-1 antibody, Pembrolizumab (initially called Lambrolizumab, developed by Merck) also entered clinical trial and showed essentially identical results [
50].
Interestingly, the phase I trials with Nivolumab and Lambrolizumab and with the anti-PD-L1 antibody, BMS 936559, were not mature enough for median or overall survival data, the results of the first rounds of PD-1 as well as PD-L1 targeted trials nonetheless established a number of facts. First, it was quite clear that all three reagents have anti-tumor activities (the response rates against melanoma for Nivolumab, Lambrolizumab, and BMS 936559 were 28, 38, and 17%, respectively); second, Nivolumab and the anti-PD-L1 antibody exhibited varied degrees of clinical activities in tumors other than melanomas such as in lung cancer, renal cancer, and ovarian cancer); third, the reagents targeting anti-PD-1 were found to be effective in patients that have previously received Ipilimumab; and fourth, the toxicities of all three reagents seemed to be considerably less severe [
48‐
50].
Subsequently, analyses of multiple phase III trials with anti-PD-1 antibodies involving much larger patient cohorts confirmed that not only both reagents showed anti-tumor activities, both anti-PD-1 antibodies prolonged median survivals (16.8 months for Nivolumab and 31 months for Pembrolizumab) as well as prolonged overall survival rates (43% for Nivolumab and 49% for Pembrolizumab at 2 years) [
51,
52]. Importantly, in line with the Kaplan -Meier survival curve for melanoma patients receiving Ipilimumab flattening at the 3-year mark, the survival curves with the anti-PD-1 targeted therapies were also found to flatten at the 3-year mark [
51,
52
]. Similarly, a recent extended follow-up of the Nivolumab trial [
53], presented at the American Association of Cancer Research Annual Meeting 2016 at New Orleans, showed the overall survival curve plateauing at 4 years with 34% alive at 5 years.
A head to head comparison between the CTLA-4 blocking agent, Ipilimumab, and the PD-1 blocking agent, Pembrolizumab, has been carried out [
54]. When the progression-free survival (PFS) and overall survival with the two were compared, the median estimates of PFS of patients receiving Pembrolizumab10 mg/kg every two weeks was found to be somewhat longer than that with Ipilimumab (5.5 vs. 2.8 months). The overall survival at 1 year was also significantly better with Pembrolizumab (74.1 vs. 58.2%, respectively). Similarly, the combination of Ipilimumab and Nivolumab has also been tested [
55‐
57]. A phase I trial including nearly 100 patients receiving the two reagents revealed a response rate of about 40% with a 2-year survival of 79%. A subsequent randomized trial of the two compared with Ipilimumab alone showed a much better response rate with the combination (61% for the combination vs. 11% for Ipilimumab alone [
57]. Finally, when a combination of Ipilimumab and the standard chemotherapeutic agent for melanoma, Dacarbazine, was compared with Dacarbazine and placebo alone, a significant improvement in overall survival of previously untreated melanoma patients receiving the combination was observed [
58].