Cancer immunotherapy: potential involvement of mediators

The description of a cell-free soluble anti-tumour factor by Carswell et al. in 1975 (Proc Natl Acad Sci USA, 72: 3666–3670) was followed by a long series of experimental and clinical investigations into the role of cell-free mediators in cancer immunotherapy. These investigations included research on the effects of macrophage–derived eicosanoids (cycloxygenase and lipoxygenase derivates of arachidonic acid) and of monokines such as tumour necrosis factor-α, interleukin-1 and granulocyte–monocyte–macrophage–colony stimulating factor) and of lymphocyte products: interleukins and interferons. The investigations yielded information on the effects of various factors on macrophage and T-cell activation in vitro, determination of direct anti-tumour properties on animal and human tumour cells in vitro and on therapeutic effectiveness in tumour-bearing individuals either alone or in combination with other therapeutic factors and their production by tumour cells. During recent years much effort has been dedicated towards the use of the tumour cells transfected with cytokine genes in the preparation of cancer vaccines. Cycloxygenase products (prostaglandins) were usually assumed to inhibit expression of anti-tumour activity by macrophages and an increase in their production in cancer patients was considered as a poor prognostic index. Lipoxygenase (leukotrienes) products were assumed to exhibit antitumour activity and to induce production of IL-1 by macrophages. Interleukins 2, 4, 6, 7, 12 and the interferons were extensively tested for their therapeutic effectiveness in experimental tumour models and in cancer clinical trials. The general conclusion on the use of cell-free mediators for cancer immunotherapy is that much still has to be done in order to assure effective and reproducible therapeutic effectiveness for routine use in the treatment of human neoplasia.


Introduction
The eld of cancer immunotherapy began approximately 100 years ago with rather 'naive' attempts to use anti-tumour antibodies raised in various animals for treatment in human sarcoma patients. 1,2 At around the same time Coley was probably the rst (1891, cited in reference 3) to suggest that cell-free ltrates of bacteria might posses anti-tumour activities.
Macrophages are known to produce and secrete derivates of arachidonic acid and cytokines (such as TNF-a and Interleukin-1 (IL-1). T and B cells produced and released a long series of interleukins; both macrophage and T and B-cell products are involved in the interaction between the immune system and tumour cells in a number of ways: direct anti-tumour activity, activation of immunocompetent cells in vivo and/or in vitro, and changes in their production in vivo in tumour-bearing hosts. Finally, it was shown that tumour cells themselves might secrete some of the above mentioned products and develop a means to resist the antitumour activity of various biologically active products.
The data accumulated on the activities of macrophages and T-cell products against tumour cells helped to devise certain immunotherapeutic protocols rst in experimental animal tumour systems and afterwards in human neoplasia. The immunotherapeutic protocols were based on either treatment with cell-free products alone or in combination with other treatments. They also helped in devising ways to activate in vitro immuno-competent cells for therapeutic uses in vivo.
The aim of this review is to discuss the potential involvement of cell-free mediators in cancer immunotherapy. On reviewing their antitumour activities it should be mentioned that some of these mediators are also involved in in ammatory processes and that their production is closely interelated.

Eicosanoids and Cancer
Cycloxygenase (prostaglandins) and lipoxygenase products have been used in experimental tumour systems and are involved in human neoplasia.

Prostaglandins
It has generally been assumed that prostaglandins (especially PGE 2 ) inhibit anti-tumour activity of macrophages. Thus a PGE 2 inhibitor, indomethacin, enhanced the macrophage cytostatic activity in vitro against MOPC-315 murine plasmacytoma cells. 4 Indomethacin stimulation of macrophage cytostasis was inhibited by PGE 2 5,6 and this enhancement of anti-tumour activity was also reported in vivo against Ehrlich murine tumour cells. 7 -10 The effect of endogenous and exogenous prostaglandins on macrophage functions was described: culture conditions that caused increased PGE 2 production by activated macrophages resulted in an inhibition of their tumoricidal activity whereas production of high levels of PGE 2 by resident and elicited macrophages was associated with an increase in their tumoricidal activity. 11 In another work, 12 it was reported that a subcutaneous injection of polyacrylamide beads in mice induced a population of immature macrophages which became fully cytostatic to syngeneic P815 plasmacytoma when stimulated in vitro by LPS. Blocking of PGE synthesis by indomethacin prevented the effect of LPS and addition of PGE 2 did not reverse the indomethacin effect but inhibited the macrophage-mediated cytostatic activity. 12 Suppression of macrophagemediated tumour cytotoxicity was correlated with an increase in secretion of prostaglandin from the macrophages of breast cancer patients. 13,14 Elevated prostaglandin production in human breast cancer was considered a marker of high metastatic potential for neoplastic cells, the increase in PG production occurred early in the course of breast cancer and decreased later in the course of tumour development. 15 It seems that PGE 2 production by human monocytes is by a subset of cells other than the cells which produce IL-1. 16 It should also be mentioned that cancer cells also produce PGE 2 and in this context it was reported that the amount of PGE 2 released by cancer cells which metastasized into the liver of tumour-bearing rats was higher than that of cells metastasizing into the kidney. 17 An increase in prostaglandin levels in cancer patients was also reported: 18 of plasma prostaglandin F levels in cases of tumours of the female genital tract and of plasma 6-oxoprostaglandin F 1a in cases of gynecological tumours. 19,20 On the assumption that prostaglandins can suppress the development and expression of effector cells, clinical studies were initiated to determine the effect of piroxicam (a prostaglandin antagonist) in patients with recurrent unresectable squamous cell carcinoma or lymphoepithelial carcinoma of the head and neck. 21 Although some improvement in immune reactivity was noted, more studies to correlate the improvement in immune reactivity with therapeutic effectiveness of either prostaglandin antagonists alone or in combination with other treatments are needed. 21 -23 A paradoxical effect of indomethacin on lymphokine-activated killer cell (LAK) activity in cancer patients was described: indomethacin enhanced LAK activity in patients with no distant metastases but depressed LAK activity in patients with such metastases. 24 Apparently, these effects of indomethacin are not related to the PGE inhibiting property of this compound. 24 Increased synthesis of prostaglandin by macrophages from breast cancer patients was assumed to inhibit macrophage mediated cytotoxicity. 25

Leukotrienes
Leukotrienes (lipoxygenase pathway of arachidonic acid) are usually considered to enhance the potential of the immune response. 26,27 Thus, leukotriene and indomethacin enhance additively the macrophage anti-tumour cytostatic function. 4,28 Leukotriene B4 (LTB4) was reported to augment human monocyte cytotoxic activity and enhance monocyte production of hydrogen peroxide, IL-1 and TNF. 26 5-Lipoxygenase activation was also described to facilitate an IL-1 transduction signal. 29 Augmentation by leukotrienes of IL-1 production by human monocytes was also described in other reports. 30,31 Lipoxygenase speci cally inhibited indomethacin stimulation of anti-tumour macrophage cytostasis 32 and reversed macrophage cytostasis, induced by the calcium ionophore A23187, towards P815 tumour cells in vitro. 31 Leukotriene C4 was reported to be an essential 5-lipoxygenase intermediate in A23187-induced macrophage cytostatic activity against P815 tumour cells. 33 Products of the lipoxygenase pathway were involved in human natural killer cell cytoxicity. 34

Cytokines
Immunocompetent cells reported to produce and secrete cytokines involved in the reaction of the organism to tumour cells were macrophages, T and B cells.

Macrophage-derived cytokines
Biologically active products described as being involved in the interaction with tumour cells were tumour necrosis factor-a (TNF-a) interleukin-1 (IL-1) and granulocyte ±macrophage colony stimulating-factor (GM-CSF).

TNF-a
The rst description of TNF-a as a macrophage product was made by Carswell et al. 35 Since then, several reviews have appeared which have been concerned with the characterization and properties of this cytokine. 3,36 -38 The mechanism and activity spectrum of TNF-a has been the topic of several investigations: it was reported that exposure of human cervical carcinoma cells to Concanavlin A (ConA) increased the total number of binding sites for rTNF-a but blocked the transduction of the signal for the cytotoxic response. 39 MethA sarcoma cells were found to be sensitive in vitro to human tumour recombinant TNF-a and expressed low numbers of TNF-a receptors. 40 TNF-a (unlike INF-c or IL-1a), induced regression of subcutaneous MethA implants. 40 It was assumed that the primary lesion induced by TNF-a is vascular and the mechanism(s) involved in generation of speci c cell-mediated anti-tumour immunity induced by the TNF-a treatment was not clear. 41 It was claimed that there are multiple pathways leading to resistance to TNF-a induced tumour cell cytotoxicity, among them production of transforming growth factors by tumour cells and ampli ed expression of certain oncogenes. 41 Apparently, viable activated monocytes produce other lytic factors in addition to TNF-a and IL-1 because TNF-a and IL-1 associated with plasma membranes of activated human monocytes lyse only monokine-sensitive tumour cells whereas viable activated monocytes lyse both monokinesensitive and monokine-resistant tumour cells. 42 The killing of tumour cells by TNF-a was assumed to involve internalization of this ligand in the target cells. 43 The anti-tumour effect of TNF-a and of Interferon-c against MmB16 murine melanoma was potentiated by macrophage colony-stimulating factor. 44 The anti-tumour activity of recombinant human TNF-SAM1 was enhanced by connecting the TNF compound to thymosin b4. 45 TNFa acted synergistically with IL-1 in inhibiting the growth of A375 cells. 46 An experimental study carried out in nude mice showed that intratumoral injection of an adenoviral vector containing radiation inducible DNA sequence of the Egr-1 promoter linked to a cDNA encoding TNF-a (Ad.Egr-TNF) enhanced the tumoricidal action of ionizing radiation in a human epidermoid carcinoma xenograft. 47 By another experimental approach it was shown that murine tumour cells transduced with the gene for TNF-a, regressed unlike non-transduced tumour cells after an initial phase of tumour growth. 48 The results in vitro and in vivo from experimental tumour models which indicated the antitumour activity of TNF-a, prompted initiation of clinical trials devised to determine the therapeutic effectiveness of TNF-a in human neoplasia. Intravenous injection of TNF-a in 18 cancer patients led to some clinical improvement in three lymphoma patients. 49 Recombinant TNF-a was given in combination with TNF-c in phase I trials in 36 patients with solid tumours. Sideeffects such as fatigue, fever and chills occurred: in one patient with melanoma there was a mixed response and in one patient with mesothelioma there was transient clearance of ascites from malignant cells. 50 Disappointing results were reported in a phase II study of recombinant human TNF-a in 127 cancer patients. The conclusion of the authors was that: ''rhuTNFa does not appear to have signi cant anti-tumour activity''. 51 A similar conclusion was made in a phase II trial of rTNF-a in 22 patients with adenocarcinoma of the pancreas: ''No objective responses were observed''. 52 A phase II study of recombinant TNF-a was also performed in a group of 26 renal cell carcinoma patients. The conclusion was: ''rTNF given as described, has only modest anti-tumour activity in renal carcinoma and produces considerable toxicity. We plan no further studies of rTNF in this disease''. 53 In a phase I study including 16 evaluable patients with various types of metastatic cancer, there was evidence for antitumour effect in two patients. 54 The property of TNF-a as an immunomodulator was reported: 55 pretreatment of monocytes with IFN-a, IFN-c, IL-1 or TNF-a resulted in enhanced human monocyte toxicity.

Interleukin-1
The production, characterization and properties of IL-1 have been summarized in several reviews. 56,57 It was stated that IL-1 is a key mediator of host response to microbial invasion and that it acts as a true hormone produced during infection and in ammation. 57 Human recombinant IL-1 (hrIL-1) induced proliferative responses of T cells in the presence of suboptimal concentrations of mitogen and doubled the response to higher concentrations. 57 Human recombinant IL-1 induced release of IL-2 by T cells and acted as a potent in ammatory agent by inducing dermal broblast PGE 2 production in vitro and of fever in rabbits and mice. 57 The overall conclusion from these data was that IL-1 possesses both immunological and in ammatory properties. 57 Human Interleukin-1 acted as a cytocidal factor for several tumour cell lines 58 and promoted human monocyte-mediated tumour cytotoxicity. 59 Human monocytes stimulated with pneumococcal cell surface components produced IL-1 but not TNF, thus showing independence of production of the two cytokines. 60 As mentioned already 46 IL-1 acted synergistically with TNF-a against tumour cells. IL-1a enhanced carboplatinum antitumour activity against human ovarian cells in vitro and in vivo. 61 In an in vitro study with human melanoma cell lines it was shown that tumour cells which secrete IL-1 exhibited increased adhesion to endothelial cells. 62

Gr anulocyte ±m acroph age colony stimul ating f acto r (GM-CSF)
The effect of intravenous and intraperitoneal administration of GM-CSF was examined in a phase I trial of 13 cancer patients refractory to standard chemotherapy. 63 Administration of MG-CSF was well tolerated but no data were provided on clinical improvement. 63 In another study, it was reported that administration of GM-CSF in 24 patients with solid tumours enhanced monocyte cytotoxicity against a human colon carcinoma line but no data are given on the effect on clinical course of the disease. 64 The partial and somehow disappointing results obtained in clinical trials with TNF-a may be due to several factors: · TNF-a can be administered only in small amounts because higher amounts are toxic and induce severe side effects.
· TNF-a is a relatively small molecule and as such, is rapidly cleared after injection.
· It is possible that some human cells are resistant to TNF-a or develop mechanism(s) of defence against the anti-tumour activity of TNF-a.
· Anti-tumour activity in vivo requires more than one anti-tumour factor.
· TNF-a in vivo does not occur suf cient concentration in its contact with tumour cells.
In view of the results obtained in clinical trials with TNF-a, experiments were undertaken to use activated human macrophages or activated human peripheral blood for therapy. It was assumed that macrophages can act indiscriminately against immunogenic or nonimmunogenic tumours, and that they might release monokines continuously in vivo (in addition to TNF-a and/or IL-1) which would be active against the tumour cells. It should be mentioned in this context that activated macrophages might release anti-tumour cytostatic products unrelated to IL-1, TNF-a and INF-a/b. 65 Results of in vitro experiments showed that peritoneal human macrophages obtained from renal patients on continuous ambulatory peritoneal dialysis (CAPD) can be activated in vitro by LPS to express anti-tumour activity and as such they acted in vivo against a human tumour implanted subcutaneously in nude mice. 66 Similarly, human peripheral blood cells from cancer patients, activated in vitro by IFN-c and LPS, reacted against a human tumour growing in nude mice. 67 These results prompted clinical trials with autologous human peripheral blood monocytes activated in vitro and reinjected in to the cancer patients. 68,69 The therapeutic effectiveness of this procedure was limited 68,69 and probably more clinical trials are required in order to improve conditions for therapy with activated macrophages.

Interleukin-2
Interleukin-2 (IL-2) is one of the main products of T cells and was extensively studied in the context of its anti-tumour activity. IL-2 was reported to increase human natural killer (NK) cell activity in vitro and this activity was partially reduced by monocytes due to PGE 2 production. 70 Murine lymphocytes cultured in the presence of IL-2 lysed syngeneic murine tumour cells. 71 The lytic activity was attributed to the occurrence of lymphocyte activated killer (LAK) cells Thy-1 + Lyt -1-2+. 71 The described effects of IL-2 in enhancing anti-tumour activity promoted a series of clinical trials devised to determine the therapeutic effects of IL-2 in cancer patients. It was found that of 106 patients with metastatic cancer receiving LAK cells plus IL-2, eight had complete responses, 15 had partial responses and ten had minor responses. 72 The same group of researchers reported that autologous tumour in ltrating lymphocytes (TIL), cultured in the presence of IL-2, lysed melanoma tumour cells. 73 Melanomaspeci c cytolytic tumour lymphocytes derived from TIL grown in the presence of IL-2, and injected together with IL-2, induced tumour rejection in vivo possibly by reacting against the gp100 epitope. 74 A phase I trial in a total of 31 evaluable patients with metastatic cancer of the breast, gastric cancer, colorectal cancer, melanoma, non-small cell lung cancer, osteosarcoma or renal cancer, received a combined treatment of IL-2, followed by TNF-a and indomethacin: 75 two partial responses were seen (in breast and renal cancer). 75 In another clinical trial, 16 patients with advanced renal cell cancer (stage IV) received a combined treatment of cyclophosphamide, INF-a and IL-2. 76 Two patients had a partial response, two had a minor response and three patients achieved stable disease. 76 The effect of IL-2 with or without LAK cells was assayed in another group of 71 patients with advanced renal cell carcinoma. 77 A low level of anti-tumour response was detected and the addition of LAK cells did not improve the response. 77 IL-2 administration was found to prolong survival of some metastatic renal cell carcinoma patients with no or moderate HLA-II expression and/or no or moderate macrophage presence in the primary tumour, but was not effective in patients with both high HLA-II and high macrophage expression. 78 Some potential uses of IL-2 treatment have been described: IL-2 and IL-7 augmented the cytolytic activity and the antitumour killing spectrum of aCD3-induced activated killer cells, and such cells were suggested for immunotherapy of non-immunogenic tumours. 79 Cytotoxicity of induced LAK cells against human leukaemia was augmented in the presence of either IFN-a, IFN-c or TNF-a in cultures, and as such they might be more effective in treating human leukaemia. 80 Adoptive therapy of established pulmonary metastases with LAK cells and recombinant IL-2 has been reported. 81 Adoptive therapy with highly enriched NK cells was assumed to have a potential use in leukaemia. 82 It should be also mentioned that LAK cells are apparently not a unique cell type but a function of various types of cells. 83 Finally, one of the problems of IL-2 therapy is the toxicity of the compound: in this context, it was suggested that the oxygen freeradical scavenger, dimethythiourea, ameloriates pulmonary permeability and vascular leak syndrome associated with multiple-dose IL-2 therapy without inhibiting IL-2 induced anti-tumour cytotoxicity. 84 On the other hand, the tumourassociated antigen 90K and the soluble IL-2 receptor were associated with poor prognosis in human ovarian cancer. 85 Another way of promoting the anti-tumour activity of IL-2 was by gene transfer into tumour cells. 86 -89 The use of a human renal carcinoma line transfected with IL-2 and/or INF-a gene has been suggested for the preparation of live cancer vaccines. 90 Combined treatment with granulocyte colony stimulating factor and IL-2 increased the survival time of nude mice bearing human ovarian cancer cells. 91

Interleukin-4
Interleukin-4 was rst described as B-cell stimulatory factor. IL-4 was reported to inhibit tumour growth of syngeneic mammary adenocarcinoma, plasmacytoma, 92 and renal carcinoma 93 by using cytokine gene transfer into the murine tumours. Repeated injections of small amounts of IL-4 around the tumour draining nodes of mice resulted in growth inhibition of poorly immunogenic and non-immunogenic tumours (cited in reference 94). Transfection of IL-4 into tumour cells induced release of this cytokine and it was effective, in vivo, against a wide range of tumour cells implanted in nude mice. 92 Similarly, treatment of established murine renal cancer by tumour cells engineered to secrete IL-4, induced speci c T-cell dependent systemic immunity against the non-transfected tumour. 93 To my knowledge, IL-4 therapy has not yet been tested in human neoplasia.

Interleukin-6 (INF-b 2 )
Interleukin-6 is induced in T cells by antigen stimulation. Puri ed human recombinant (rIL-6) mediated substantial reduction in the number of tumours. 95 Recombinant human IL-6 produced in Escherichi a coli inhibited the growth of human breast carcinoma and leukaemia/lymphoma cell lines in vitro. 96 Anti IL-6 receptor antibody prevented muscle atrophy in Colon-26 adenocarcinoma bearing mice, thus suggesting that this antibody could be a potential agent against muscle atrophia in cancer cachexia. 97 IL-6 gene transfer into murine syngeneic tumours inhibited growth of lung carcinoma 98 and of sarcoma. 99 IL-6 gene transfected into Lewis Lung Carcinoma tumour cells suppressed the malignant phenotype and was effective against parental metastatic cells. 98 Mice rejecting murine brosarcoma cells, transduced with retroviral vectors containing the murine IL-6 gene and secreting IL-6, exhibited a later resistance to challenge with wild tumour cells. 98 Aquisition of the ability to synthesize endogenous IL-6 markedly accelerated the growth of weakly tumorigenic rat urothelial cells, but did not induce a tumorigenic phenotype in non-tumorigenic cells. 100

Interleukin-7 (B/ T m atur ation f acto r)
Murine plasmacytoma cells transfected with the IL-7 gene produced IL-7 in vivo and were completely rejected in syngeneic mice by a Tcell dependent process. 101

Interleukin-12
IL-12 has received considerable attention during recent years as a strong anti-tumour agent: IL-12 was reported to react in vivo against B16F10 tumour-bearing mice and its anti-tumour effect was inhibited by anti-IFN-c antibody; 102 IL-12transfected broblast cells admixed with the murine melanoma, BL-6, showed that local IL-12 expression suppressed tumour growth and promoted development of speci c anti-tumour immunity; 103 IL-12 engineered dendritic cells were effective in vivo against murine tumour cells. 104 In another report, it was shown that anti IFN-c antibody blocked IL-12-mediated tumour regression in mice. 105 In view of the results obtained from IL-12 therapy against murine tumours, clinical trials were initiated to determine the effect of IL-12 in human cancer. Phase I clinical trials of IL-12 gene therapy were done by direct injection of tumours with genetically engineered autologous broblasts. 106,107 Out of 13 cancer patients (six breast, ve melanoma, and two head and neck), signi cant reduction in tumour size was observed in three patients with melanoma and one with head and neck cancer. 108

Interferon-a 2b
A clinical trial was done in renal cancer and melanoma patients by treatment with INF-a 2b . Of 12 melanoma patients, four patients showed a partial response whereas eight patients progressed. 109 In the case of 35 patients with advanced renal cancer, an increase in immune response potential was observed. 109 The conclusion was that more studies are required to determine the therapeutic effectiveness of INFa 2b . 109 A human renal carcinoma line transfected with the IL-2 and/or the IFN a gene was suggested for use in preparation of live cancer vaccines. 90

Interferon-c (IFN-c)
IFN-c was rst described as an antiviral agent and was among the rst cytokines assayed for therapeutic effectiveness in human neoplasia. In one of the rst clinical trials done in Hodgkin's disease patients it was reported that treatment with IFN-c led to an extension of the diseasefree survival time. 110 Complete or partial remission in multiple myeloma by IFN-c treatment was also reported. 111 Remissions induced by IFN-c were also reported in patients with lymphocytic lymphoma, 112 in cases of metastatic breast cancer, 113 non-Hodgkin's lymphoma 112 and multiple myeloma. 113 However, in another study in a group of non-small cell lung cancer patients, INF-c treatment did not induce tumour regression. 114 A phase I trial on the effect of IFN-c treatment was conducted in six patients with lung cancer; 115 systemic side effects such as transient fever, nausea, headaches and u-like symptoms were noted 115 but no data on therapeutic effect were given. 115 In a recent in vitro study it was reported that human carcinoma cell lines cultured with IFN-c expressed more CD80 and CD86 costimulatory molecules and that this increase in expression was inhibited by IL-10. 116 Transfer of the IFN-c gene into murine neuroblastoma, 117 brosarcoma, 118 adenocarcinoma, 119 colon carcinoma 119 and lung carcinoma 120 cell lines induced tumour inhibition in syngeneic mice. Human peripheral blood monocytes collected from cancer patients were activated in vitro to express anti-tumour activity in the presence of IFN-c and LPS. 67,68 Anti-tumour effects vs other functions of cytokines Some facts which should be considered in the context of the anti-tumour activity of cytokines are discussed below.

Rel ationship with in amm ato ry functio ns
Macrophage-derived cytokines such as TNF-a and IL-1 are also major in ammatory mediators. 5,27,28,37,56 ,57 Human peritoneal macrophages from CAPD patients collected during infectious peritonitis are 'primed' to produce and secrete more TNF-a and IL-1 when cultured in the presence of LPS. 121,122 These data indicate that in ammation and cancer are interrelated events.

Inter actions in cy tokine production
Production and release of various eicosanoids and cytokines are closely interrelated. 37 For, example: production of IL-1, TNF-a and IL-6 by human mononuclear cells was induced by stimulatory agents such as LPS, 36,56 LPS-induced TNF-a production is inhibited by PGE 2 , 123 endotoxin, TNF-a and IL-1 induce IL-6 production in vivo; 124 production of IL1 and TNF-a was induced in human blood mononuclear cells by LPS, whereas IL-6 suppressed the induction of IL-1b and TNF-a by LPS or PHA. 125

Production of eicos ano ids and cy tokines by tumour cells
Production of eicosanoids and cytokines by tumour cells may have an in uence on the effect of these products on tumour development: tumour cells might produce PGE 2 ; 126 the response of murine tumours to indomethacin therapy was directly related to their ability to produce prostaglandin; 127 tumours from cachectic mice produced both TNF-a and IL-1a in vivo; 128 a myeloma human cell line produced both TNF-a and IL-6; 129 leukaemic cells from patients with acute myeloid leukaemia produced both IL-6 and IL-1. 130 These are a few examples; the relationship between the ability of tumour cells to produce eicosanoids or cytokines and development of cancer is not yet clear.

Production of eicos anoids and cyto kines in the tumour-be aring host
During tumour growth in rats cycloxygenase or thromboxane synthase was inhibited whereas C5 and C12-lipoxygenases of the alveolar macrophages were activated. 131 Macrophages derived from tumour-bearing animals suppressed activation of T cells, of NK cells, of LAK cells and of generation of tumoricidal activity in normal syngeneic splenic macrophages in cultures stimulated by LPS. 132 The secretion of IL-1, TNF-a but not of IL-6 was impaired in alveolar macrophages collected from tumour-bearing mice. 133 A decrease in production of IL-1 was also reported in peritoneal macrophages collected from sarcoma-bearing mice. 134 Production of IL-1 and TNF-a by tumour associated mononuclear monocytes from cancer patients was examined and showed that production IL-1 was suppressed whereas production of TNF-a was not affected. 135 LPS induced TNF-a production was impaired in macrophages from breast cancer patients, 136 but increased in patients with malig-nant brain tumours. 137 An example of correlation between production of TNF-a and PGE 2 by peripheral blood monocytes was seen in patients with bladder cancer: these patients had either higher TNF-a production or higher PGE 2 production. 138

Cell-free mediators as cancer therapeutic agents
In vitro activation of cells for immunotherapy Activation of autologous human peripheral blood monocytes by culture in the presence of LPS and IFN-c was assayed with the aim of inducing anti-tumour cytotoxic activity in the treated monocytes and to use such activated cells for immunotherapy. 68,69 Clinical trials reported limited success. 68,69 Induction of LAK cells, and later of activated tumour in ltrating lymphocytes (TIL) by culturing in the presence of IL-2 was also suggested in a series of reports. 71 -74 Clinical trials were carried out in groups of cancer patients; the results obtained indicated some therapeutic bene t in melanoma and renal carcinoma with approximately 10% clinical improvement in patients treated with both IL-2 and LAK cells. 72

Direct administration of cytokines in cancer patients
As previously mentioned the results of treatment with TNF-a were disappointing; injection of IL-2 with or without LAK cells was problematic due to the toxicity of the agent; the effect of IL-2 in patients with squamous cell carcinoma of the head and neck was limited to temporary regression of the tumours. 139 Another approach suggested was to counteract the activity of PGE 2 by using a PGE inhibitor: while some improvement in immunological functions was observed no data are yet available on the therapeutic effectiveness of this treatment. 21

Transfection of tumour cells with cytokine genes
This approach has been extensively investigated during recent years, rstly in experimental tumour models and later in clinical trials. The transfected tumours released the appropriate cytokine in vitro and in some cases induced Tcell mediated speci c anti-tumour immunity. Partial success was observed in terms of tumour regression and clinical improvement. 140 -142 Productive transfer of the IL-2 gene was shown for melanoma, renal cell carcinoma, neuroblastoma and acute leukaemia cell lines. 86,108,139 -141 Several reviews have been published on the role of cytokines in cancer therapy: in a review published in 1989 it was concluded ''The clinical results obtained with cytokines (INF-a, IL-2, TNF-a, GM-CSF, G-CSF) expected to show direct tumoristatic or tumoricidal effects have been disappointing''. 143 A more optimistic view was expressed in another review: ''local presence of cytokines, either injected repeatedly at the tumour site or released by cytokine-engineered tumour cells, arouses immunogenicity in apparently nonimmunogenic spontaneous tumours'' and as such they might indicate ''potential use of cytokines as a component of new tumour vaccines''. 144 In another review, it was concluded: ''Clinical applications [of cytokines] are progressing, but many trials must follow to assess precisely the multitude of potential uses of these molecules''. 94 Finally, in a recent review it was concluded: ''Although several years will probably be required to establish the true impact of the gene-transfer modalities . . . technological advances have opened prospects for the management of cancer patients''. 142

Concluding Remarks
The 'golden' era on the role of cytokines in the ght against cancer started with the description by Carswell in 1975 of tumour necrosis factor. 35 Since then, various aspects of the interaction between eicosanoids, macrophage and lymphocyte cytokines and tumour state have been described and investigated. The various topics of investigation included: · therapeutic effectiveness of tumour cells transfected with cytokine genes. It is assumed that transfected tumour cells are more immunogenic and as such they induce a T cell mediated response against the wild type non-transfected tumour. They might also secrete cytokines in vivo which react against the tumour cells.
Marked progress has been made during recent years on the interaction between macrophage and lymphocyte products and tumour cells, and on their possible role in immunotherapy against cancer. However, many more clinical trials are required before these agents can be used in routine therapy of human neoplasia.