Lessons from Coley's Toxin

1. Coley's Toxin 

Despite the fact that surgeon William B. Coley's pioneering work in cancer immunotherapy with bacterial toxins was carried out over a century ago [1], the significant clinical observations [2] and the mechanism behind it remain unexplained until now. A compiling of Dr. Coley's clinical observations indicated that in certain tumor types such as soft tissue sarcoma and lymphoma, the response is remarkable even by today's standards. For example, out of 104 inoperable sarcoma patients who were treated with Coley's Toxin alone, over 50% demonstrated complete tumor regression at the time of therapy and subsequently more than 5-year survival. Further, approximately 20% of patients were cured and survived over 20 years [3]. These response rates remain unsurpassed by subsequent and even recent modern immunotherapy trials. However, when other types of cancer, usually carcinomas, were treated, the responses were generally not as good [2], [3]. Why did a therapy, which we know today works by activating the immune system, work well in some patients, but not others? In order to know why Coley's Toxin failed in some patients, we first need to know why it worked in some others. We then can understand what makes a patient a responder to Coley's Toxin and the biological principle it represents. Clearly the answers to these questions demand a better understanding of the biological factor(s) of Coley's Toxin.

2. Tumor necrosis factor (TNF) 

Knowing that the major effect of Coley's Toxin comes primarily from endotoxin (LPS) of Gram-negative bacteria [4], [5], the discovery of TNF in the 1970s [6] was thought to have provided a satisfactory answer [3], [7]. However, subsequent testing using recombinant TNF showed many of the toxicities of Coley's Toxin/endotoxin without equaling the significant anti-tumor efficacy in both clinical trials and laboratory studies. It is worth noting that even amid the enthusiastic belief that TNF had provided all of the answers to Coley's Toxin, some scientific challenges also appeared. North and colleagues demonstrated that in the absence of T cells, endotoxin can only induce transient tumor necrosis, but not permanent regression, thus it cannot be singly responsible for the entire anti-tumor activity of endotoxin/Coley's toxin [8]. Not only are T cells required for endotoxin-induced tumor rejection, these T cells must be sensitized to the incipient tumor before beginning endotoxin therapy in order for complete tumor rejection to occur [9].

Although TNF failed to materialize the anti-tumor dream of cancer immunologists, the science and technology associated with its identification and production supplied waves of subsequent cytokines. Almost all of these cytokines have been tested in animal tumor models, but none has demonstrated similar anti-tumor effects of endotoxin, except one: interleukin-12 (IL-12).

3. IL-12 

IL-12 was first described as a NK cell activating cytokine, and subsequent studies showed IL-12 also target T cells. It is now well recognized that IL-12 plays a central role in the connection between innate and adaptive immunity by promoting the development of cell-mediated Th1 type of immunity. Corresponding to the ability of IL-12 to target both T and NK cells, the anti-tumor activities of IL-12 from various studies can also be divided into two types: the one that depends on NK/NKT and the one that depends on classic T cells. NK/NKT-mediated anti-tumor response activated by IL-12 is often seen in models of non-established tumors in which it destroys the development of freshly inoculated tumors [10]. In these models tumor cells are killed through a perforin-dependent effector mechanism similar to that of NK and cytotoxic T cells. In contrast, the T cell-mediated anti-tumor response induced by IL-12 is highly effective in established tumor models [11], [12], [13], [14]. Although the anti-tumor effect by this type of response can be dramatic eradicating late-stage large tumor burdens completely, it is not broadly observed in many tumor models, but only in so-called immunogenic tumor models [14]. It is dependent on interferon-gamma, but not NK/NKT cells [14], [15]. Unlike the NK/NKT-mediated response, T cell-mediated tumor rejection by IL-12 does not require perforin and may involve activated macrophages [16] and other unique effector mechanisms for tumor cell destruction.

Several unique characteristics are shared by endotoxin- and IL-12-induced antitumor responses: (1) All responders to both therapies are immunogenic tumors [8], [14]. (2) Both therapies prefer “established” tumors as rejection of palpable tumors established 7–10 days is more effective than non-established tumors implanted 2–3 days [8], [17]. Both responses require interferon-gamma [13], [18]. (4) Both responses require the so-called pre-existing immunity [9], [14], which is a spontaneous T cell response to the incipient tumor upon tumor establishment [19]. In addition, we have observed a strict parallel in tumor models responding to both LPS and IL-12. For example, Sa1 sarcoma used by North and colleagues in many of their endotoxin-mediated anti-tumor studies were found to respond to IL-12 equally well [14], [16]. On the other hand, the MCA207 sarcoma that we and others have used in tumor rejection models with IL-12 was found to respond to endotoxin as well, albeit that IL-12 always seems more effective than LPS (Table 1). A direct link between endotoxin and IL-12 is suggested by the finding that certain host immune cells such as macrophages and dendritic cells stimulated with LPS in vitro produce large amount of IL-12 [20], [21]. Evidence also suggests a link in vivo as in the absence of IL-12, endotoxin administered in vivo fails to induce interferon-gamma production and septic shock [22]. Giving the close similarity of the antitumor characteristics between endotoxin and IL-12, and the fact that endotoxin is able to induce IL-12 production, we hypothesized that endotoxin exerts its antitumor effect through the induction and the biological activity of IL-12. If so, it is predicted that endotoxin-induced T cell-dependent rejection of established tumors would be abolished in the absence of IL-12. This is indeed the case as the evidence from experiments using IL-12 gene knockout mice has indicated (Table 1). Since exogenous IL-12 alone was able to induce tumor rejection in IL-12 gene knockout mice where endotoxin failed (Table 1), IL-12 is not only necessary, but also sufficient, for endotoxin-induced tumor rejection. Thus, based on the evidence, it is reasonable to believe that IL-12 is ultimately responsible for the anti-tumor effects of Coley's Toxin.

Table 1. IL-12 is necessary for endotoxin-induced rejection of established tumors in mice

	Host	Treatment	Cure-rate (%)
	Normal	None	0
	Normal	IL-12	100
	Normal	LPS	80
	IL-12 knock out	None	0
	IL-12 knock out	IL-12	100
	IL-12 knock out	LPS	0
	IFN-γ knock out	None	0
	IFN-γ knock out	Il-12	0
	IFN-γ knock out	LPS	0

8-day subcutaneously established (2–4mm) murine MCA207 sarcoma tumors grown in normal, IL-12 (p40) or Interferon-gamma gene knockout (KO) mice were treated with i.p injection of saline (none), recombinant murine IL-12 (IL-12, 0.5μg, day 8,10 and 12) or Escherichia coli LPS (endotoxin, 75μg, day 8 and15). Cure represents the percent of mice with durable complete tumor regression.

4. Pre-existing immunity 

Now that we know that IL-12 is responsible for the effect of Coley's Toxin against established large tumors, new knowledge obtained from the study of IL-12-induced tumor rejection may shed light on the question why Coley's Toxin worked well in some cases, but not others. In the case of IL-12- (and endotoxin)-induced tumor rejection, an anti-tumor immune recognition in the form of a Th1 response must be present prior to the start of therapy in order for the host to respond to the therapy [14], [15]. This requirement, which was called “pre-existing immunity” by North and colleagues [9], comes from the fact that IL-12 receptors are preferentially expressed on activated, but not naïve or resting, T cells. Thus only tumor-sensitized T cells are able to respond to IL-12. It explains the paradoxical observation that IL-12 therapy is more effective against established, than freshly inoculated, tumors and the fact that all high responders to IL-12 therapy are immunogenic tumors. Extending this principle to clinical observations by Coley, we hypothesize that the high responders to Coley's Toxin were patients who had pre-existing immunity. Since sarcoma and lymphoma patients responded best to Coley's Toxin [3], we expect to find a significant number of these patients who may be responders to IL-12 therapy as well. Unfortunately, Coley's early work has also shown that most major cancers did not respond well to his toxin therapy, thus indicating that many of these cancers are probably weakly or non-immunogenic. It is generally observed that most spontaneously arising tumors in the mouse are also non-immunogenic. In these cases, either T cells are not yet sensitized to the incipient tumor or the sensitized T cells are suppressed by immune regulatory T cells called suppressor T cells. At least one specific group of CD4+CD25+T cells has been identified as suppressor cells of the anti-tumor response, as removing these cells can release the hidden pre-existing immunity [23], [24], [25].The challenge, therefore, is to find ways to induce a decent Th1 immune recognition in hosts bearing non-immunogenic tumors if IL-12 therapy is going to be effective in a broad range of patients.

5. Reasons for prior failures 

Previous clinical trials with IL-12 have not yielded success and the reasons for failure may be multiple. First of all, melanoma and renal cell carcinoma patients were selected in these trials. Coley had indicated that melanoma (“melanotic sarcoma” in his writing), is the type of cancer “with which I have had no permanent success”. Secondly, the early trials have used intensive dosing of IL-12 favoring the generation of NK/NKT, but not T cells with high toxicity. Thirdly, IL-12 was used alone without manipulation to remove regulatory T suppressor cells. Since our animal study results have demonstrated a dramatic difference in large tumor burdens with the addition of cyclophosphamide [13], [14], [16], it is possible in those previous trials that even when a patient had pre-existing immunity, it is likely to be under suppression. Thus the anti-tumor T cells may not respond to IL-12 due to a lack of IL-12 receptor expression.

Based on our new understanding described here, future clinical trials should take these factors into consideration. For example, for direct IL-12 therapy, we should try to select patients who possess as much pre-existing immunity as possible. Although methods and standards for selecting and measuring pre-existing immunity in humans are yet to be established, various recent studies have provided evidence for the presence of spontaneously generated anti-tumor immunity that appears to provide a better prognosis [26], [27], [28], [29], [30], [31]. Additionally, the use of drugs or antibodies to remove suppressor cells should be helpful, although the timing of these agents needs to be carefully worked out as they also remove activated effectors cells bearing the same cellular characteristics (e.g. cycling) or cell surface marker (CD25). Finally, we need to think of strategies to combine efforts to raise pre-existing immunity in patients and then subject those who show robust signs of immune response to subsequent IL-12 therapy. Previous clinical studies have shown the generation of detectable anti-tumor T cell responses in patients without correlation with complete tumor eradication [32]. In light of the pre-existing immunity concept, we may think of this situation similar to a case of implanting an immunogenic tumor to a host. The host shows clear signs of immune responses against the tumor, yet the host still dies of the cancer. Tumor vaccine and other approaches may help us to get to the stage of pre-existing immunity, but not necessarily to tumor regression. Subsequent IL-12 therapy will do that. Without subsequent IL-12, it is like planting a seed without subsequent nourishment. On the other hand, giving IL-12 bluntly to all patients is like pouring nourishments on a vacant land. In this respect, better response rate and efficacy are expected from future IL-12 clinical trials in patients with pre-existing immunity either spontaneously generated or created through deliberate immunological manipulations.