Isaacs and Lindenmann discovered interferon in 1957 while searching for a substance that blocked viral infection of cells.[1] The medium from cultures of chicken cells, infected with influenza virus, was added to other cultures of chicken cells. This medium prevented the infection of the second cultures by a different virus. The interfering substance was designated interferon. Nagano and Kojima also independently discovered this soluble antiviral protein.[2]

Interferons are a family of proteins that exhibit antiviral activity.[3] Human interferons were initially categorised into leucocyte, fibroblast and immune interferons on the basis of their cellular origin.[3,4] Interferons are divided into two types based on their ability to bind a common receptor. Type I interferons are acid stable and are the ligands for the type I interferon receptor; they include leucocyte or α interferon (IFNα), fibroblast or β interferon (IFNβ), τ interferon and ω interferon.[5] Type II interferon is acid labile, binds the type II interferon receptor, and is represented by immune or γ interferon (IFNγ).[5] The α interferons are glycosylated polypeptides of varying molecular weight that are produced in response to enveloped viruses, virus-infected cells, bacteria and tumour cells.[6] Fibroblasts, epithelial cells and macrophages produce IFNβ following exposure to micro-organisms.[7] IFNγ is produced by lymphocytes after antigenic or mitogenic stimulation.[7] Table I summarises the classification of interferons and their properties.

Table I
figure Tab1

Classification of interferons and their major physicochemical properties[3,4,6–9]

Full size image

1. Biology of Interferon-α

1.1 Natural Leucocyte Interferon

Natural leucocyte interferon can be produced in vitro after treating leucocytes with a virus such as Sendai or Newcastle disease virus.[4] Cantell and colleagues[10] were the first to produce partially purified IFNα. In 1979, high performance liquid chromatography was used to isolate and characterise natural leucocyte interferon; this was found to consist of at least ten species that were designated α-1, α-2, β-1 to 3, γ-1 to 5 and δ.[4,11] The same Greek letters were used a year later to formally classify interferons as α, β and γ, corresponding to the previous names of leucocyte, fibroblast and immune interferon.[12] The different molecular forms of natural leucocyte interferon were shown to exhibit varying cytostatic and immunomodulatory effects on mammalian cells.[13]

In vivo, the main producers of IFNα are the natural interferonproducing cells, specialised leucocytes producing huge amounts of IFNα in response to enveloped viruses, bacteria and tumour cells.[14] Although leucocytes are the prototypic cells producing IFNα, nearly all cells in the body can produce it.[8]

1.2 Genes, Proteins and Receptors

The genes that encode IFNα have been cloned.[15,16] To date, 14 human IFNα genes have been discovered,[4] which encode for 12 distinct IFNα proteins.[3] The cloning of a gene (IFNA2) for the human IFNα-2 protein[17,18] and the advent of recombinant genetic engineering allowed the production of sufficient IFNα-2 for clinical use. Recombinant human IFNα-2 is available in two forms, rhIFNα-2a and rhIFNα-2b.[7] In some countries rhIFNα-2c is licensed. Recombinant IFNα-2a, -2b and -2c differ in one to two amino acid positions.[9] Although rIFNα2c has been tested in clinical trials, the antitumour activities of rhIFNα-2a, -2b and -2c in vivo have not been compared with each other in randomised studies.

Interferons bind specific high-affinity receptors on the cell surface.[7] The IFNα receptor (IFNα-R) is made up of at least two polypeptide chains, IFNα-R1 and IFNα-R2 (figure 1).[5] IFNα-R1 is involved in signal transduction[19] and IFNα-R2 is the ligand-binding chain that also plays a role in signal transduction.[19] IFNα-R2 exists in three isoforms,[20] of which IFNα-R2c is the functional receptor.[21] Ligation of the receptor by IFNα produces oligomerisation of the receptor chains followed by a series of phosphorylations and initiation of a signal transduction pathway. This results in the phosphorylation of STAT (signal transducers and activators of transcription) proteins, which translocate to the nucleus as a trimeric complex, ISGF-3 (interferon-stimulated gene factor-3). ISGF-3 activates transcription of interferon-stimulated genes, with many biological effects.[21]

Fig. 1
figure 1

Signal transduction pathway for interferon-α (IFNα) signalling: (i) receptor (IFNαR) ligation and oligomerisation; (ii) transphosphorylation of Janus kinases(JaKs; JaK 1 and TyK 2); (iii) phosphorylation of the cytoplasmic tails of receptors; (iv) docking of STAT (signal transducers and activators of transcription) proteins on the cytoplasmic tails of the receptor and phosphorylation (-P) by JaKs; and (v) STAT dimers heterodimerise with p48 to form IFNα-stimulated gene factor-3 (ISGF-3), which translocates to the nucleus to activate transcription of IFNα-sensitive genes by binding to IFNα-stimulated response elements (ISRE) [reproduced from Pestka,[20] with permission].

Full size image

1.3 Mechanisms of Action and Antitumour Effects

IFNα is a pleiotropic cytokine[22] with antiviral and antitumour effects,[20] and immunomodulatory[7] and antiangiogenic activity (table II).[6,23] The complexity of its effect on cellular function is illustrated by the fact that IFNα can induce the expression of more than 300 genes.[24] The antitumour action of IFNα is mediated by the following effects.

Table II
figure Tab2

Mechanisms of action and biological properties of interferon-α[3,25,26]

Full size image
  1. (i)

    The antiviral effect of IFNα is shown by its ability to limit virally induced malignancies such as hepatitis virus-induced liver cancers and herpes simplex virus-associated Kaposi’s sarcoma.[25]

  2. (ii)

    IFNα has a number of direct antitumour effects, which include prolongation of the cell cycle of neoplastic cells,[7] i.e. a cytostatic mechanism. This is thought to be achieved by modulation of 2′,5′-oligoadenylate synthetase activity[27] or modulation of cellular oncogenes.[29] IFNα also inhibits ornithine decarboxylase,[28] an important biosynthetic enzyme, and thus may lead to depletion of metabolites required for tumour growth. IFNα may also interact synergistically with IFNγ to promote cell lysis.[30] Relatively recently, it has been demonstrated that IFNα has proapoptotic effects in human tumour cell lines, squamous cell cancers and basal cell carcinoma.[26]

  3. (iii)

    IFNα exerts indirect antitumour effects. These include upregulation of class I major histocompatibility complex (MHC) and other membrane antigens on malignant cells (e.g. melanoma tumour-specific antigens),[31] resulting in more efficient recognition/lysis of cancer cells by immune effector cells.[32] Production of antibodies to tumour cells is enhanced, resulting in increased complement-mediated and antibody-dependent cell-mediated cytotoxicity of tumour cells.[32] This may explain the augmented cytotoxicity of macrophages, natural killer (NK) cells and T cells.[33,34]

  4. (iv)

    The other important indirect mode of action of IFNα is by inhibition of angiogenesis.[35] This was observed initially in mice.[36] It was subsequently observed that recombinant IFNα therapy improved Kaposi’s sarcoma, the vascular tumour associated with HIV infection,[37] and shown that IFNα produces regression of haemangiomas, the most common benign tumour of infancy.[38–40]

1.4 Effect on Function of Cells of the Immune System

IFNα has a pivotal role in promoting T helper 1 (Th1) differentiation of naive T helper cells (Th0). From the clinical viewpoint, Th1 responses, characterised by production of interleukin (IL)-2, IFNα and tumour necrosis factor (TNF), are important because they confer protective immunity to pathogens and tumours. Th2 responses, characterised by production of IL-4, IL-5, IL-6, IL-10 and IL-13,[41] are associated with hypersensitivity reactions and autoimmune disease.[42] IFNα makes T cells more responsive to IL-12 (figure 2), thus promoting a Th1 immune response.[42] In addition to this effect, IFNα induces the generation and survival of CD8+ memory cells and is therefore important for cytotoxic T cell responses.[42] IFNα has a stimulatory effect on macrophages[8] and enhances the cytotoxic activity of NK cells.[43]

Fig. 2
figure 2

Effect of interferon-α (IFNα) in promoting T helper 1 (Th1) differentiation. Antigen (Ag) presentation by major histocompatibility complex class II (MHC II) on an antigen-presenting cell (APC) to the T cell receptor (TCR) on a naïve T cell promotes IFNα production by the APC (i). IFNα binds to its receptor (IFNαR) on the T cell (ii) and initiates the signal transduction pathway (figure 1), resulting in transcription of IFNα-stimulated genes (iii). One of the effects is upregulation of the β2 subunit of the interleukin-12 receptor (IL-12R) (iv). The high affinity receptor (IL-12Rβ1 + IL-12Rβ2) makes the T cell more responsive to IL-12 produced by the APC (v). Binding of IL-12 to its receptor (vi) activates STAT (signal transducers and activators of transcription) protein 4 (vii), which is important for IFNγ production (viii). IFNγ in turn promotes a Th1 immune response (ix) [reproduced from Tompkins,[42] with permission].

Full size image

2. History of Interferon-α in the Treatment of Solid Tumours

The initial clinical trials used crude preparations, containing <1% IFNα, isolated from buffy coats.[3] The earliest use of IFNα in human cancer was by Strander, who had treated patients with osteosarcoma since 1971. In his series, 64% of interferontreated patients were free of metastases at 2.5 years after surgery compared with 30% in a control group from other Swedish hospitals. Overall survival (OS) for Strander’ s patients was also improved at 73%, compared with 35% in controls, at 2.5 years.[44–46] After this, tumour regressions were reported in cervical cancer,[47] non-Hodgkin’s lymphoma[48] and breast cancer, myeloma, lymphoma and malignant melanoma.[49–51] In 1981, interferon was used in the neoadjuvant setting in patients with cervical cancer, and tumour regressions were again seen.[52] In 1982, a phase II study of the American Cancer Society reported a 22% response rate in breast cancer patients.[53] The numbers of patients treated in these trials were small; however, leucocyte interferon was shown to have antitumour activity. These trials were therefore very important because they provided the impetus for the development of rIFNα to provide sufficient quantities for clinical use.

The availability of large amounts of rIFNα from 1981 onwards allowed numerous phase I, II and III trials to be conducted,[54] in which its efficacy, tolerability and effect on survival in patients with solid tumours and haematological malignancies were tested. The US Food and Drug Administration (FDA) approved rIFNα for clinical use in hairy cell leukaemia in 1986.[22] Since then, approval has been given for use in other malignancies, including malignant melanoma, renal cell carcinoma (RCC) and Kaposi’s sarcoma.[22] Some tumours such as superficial bladder cancer are characterised by very high response rates; in malignant melanoma and RCC the response rate is low,[55] but these responses are extremely important because some of them are long-lasting. In addition, IFNα has been used to treat head and neck cancers, gastrointestinal tract and neuroendocrine tumours, and lung, breast, ovarian and cervical cancers (reviewed by Jonasch and Haluska,[8] John and Foon,[56] and Borden et al.[57]).

3. Malignant Melanoma

3.1 Adjuvant Therapy

Malignant melanoma accounts for 2% of all cases of cancer and 1% of all cancer deaths in the UK.[58] However, its incidence is rising and in some areas of Australia and New Zealand 1 in 30 people will develop melanoma, while in the US the life-time risk of developing melanoma has risen by more than 2-fold in the last 20 years.[59] The 5-year survival rate for patients with American Joint Committee on Classification (AJCC) stage I disease (primary tumour thickness <1.5mm) is 90%. However, for stage IIA (1.5 to 4.0mm), IIB (>4mm) and stage III (primary any size; nodal but no distant metastases) this drops to 70, 50 and 40%, respectively.[59] The reason is recurrent disease despite radical surgery.[58] Effective adjuvant treatment therefore would be expected to reduce relapses and improve survival.

rIFNα has been used widely in malignant melanoma since the 1980s. Following the publication of studies that showed response rates of about 15% in metastatic disease,[60–69] many randomised trials[70–77] have tried to establish the role of rIFNα in the adjuvant setting.[58] These are summarised in table III. They can be divided into three groups: those using high-dosage IFNα (HDIFN; ≥10MU), those using intermediate-dosage IFNα (IDIFN; 5 to 10MU) and those using low-dosage IFNα (LDIFN; ≤3MU).[78]

Table III
figure Tab3

Published randomised trials of adjuvant interferon (IFN)-α versus observation in malignant melanoma.[70–77,79–83] The overall number of patients treated in published randomised trials of adjuvant IFNα was 6323

Full size image

3.1.1 High-Dosage Trials

The first trial to show a benefit in OS for stage IIB/III disease, using HDIFN, was Eastern Cooperative Oncology Group (ECOG) EST 1684.[73] This recruited 287 patients with surgically resected high-risk melanoma. Patients were randomised to observation or HDIFN. This was given as induction (IFNα-2b 20 MU/m2/day intravenously for 5 days a week for 4 weeks) followed by maintenance (10 MU/m2/day subcutaneously thrice weekly for 48 weeks). After a 7-year follow-up, both disease-free survival (DFS) and OS were better in the IFNα arm (median OS 3.8 vs 2.8 years). The main drawbacks of HDIFN in this ‘Kirkwood regimen’ were the toxicity profile and the need for dosage reductions (~50% of patients), as two-thirdso experienced grade 3 to 4 toxicity. A retrospective analysis of quality of life concluded that the IFNα-treated group gained on average 8.9 months free of relapse (p = 0.03) and 7 months of OS (p = 0.02), against an average of 5.8 months of severe toxicity.[84] ECOG 1684 enabled the FDA to approve IFNα-2b as adjuvant therapy for stage IIB/III melanoma in 1995.[85]

The early data from ECOG EST 1684 and preliminary data from the World Health Organization (WHO)-16 trial,[70] ongoing at the time, suggested a role for HDIFN. Consequently, the intergroup trial ECOG 1690 followed.[76] This compared HDIFN with observation and with LDIFN as given in the WHO-16 study, but over 2 instead of 3 years. A preliminary analysis suggested no advantage in OS and borderline advantage in DFS (p = 0.05) for the HDIFN arm. For LDIFN there was no significant benefit in DFS or OS. This may have been due to improvements in survival of patients in the observation arm of ECOG 1690, who received IFNα salvage therapy.[58,78] Another study of HDIFN, North Central Cancer Treatment Group (NCCTG) 83–7052,[72] used IFNα at dosages as high as the Kirkwood regimen, but could not reproduce the results of ECOG 1684 or ECOG 1690. This may be related to the fact that IFNα was administered intramuscularly and only for 3 months.

Follow-up trials from ECOG 1684 and 1690 were E2696 and E1694. E2696 was a phase II trial that tested the effect of HDIFN and the GM2 ganglioside-based GMK vaccine on the induction of antibody responses. It enrolled 107 patients with resected stage IIB, III or IV melanoma into three arms: HDIFN + GMK given simultaneously (arm A) or sequentially (arm B) and GMK alone (arm C). The trial showed an improved DFS for the HDIFN + GMK vaccine arms (A and B). This finding must be viewed with caution because of the small sample size and the short follow-up time of 24 months.[79]

E1694 was a phase III intergroup study that compared HDIFN and the GMK vaccine. The National Cancer Institute’s (NCI) Drug Data Safety Monitoring Board halted this trial prematurely in May 2000 after significantly higher relapse and death rates in the vaccine arm. E1694 has now been published[80] and has reproduced the earlier findings of E1684 in demonstrating that HDIFN improves DFS and OS. Comparison of survival curves of the vaccine arm in E1694 showed this to be similar to the observation arm of E1690[86] and other adjuvant trials, suggesting that the vaccine was not worse than observation.[80]

An ongoing study, ECOG E1697, examines the role of adjuvant HDIFN for stage IIA-B patients (HDIFN, 20 MU/m2 five times a week for 4 weeks versus observation). Its aim is to establish the value of the initial induction phase with intravenous HDIFN,[87] when approximately 20% of the total yearly dose is delivered, as analysis of E1684 and 1690 appears to show that the effect of the high-dosage regimen appears to be early.[87] Another ongoing study is the Sunbelt melanoma trial, which is evaluating the role of the Kirkwood regimen in patients with histologically positive lymph nodes or reverse transcription-polymerase chain reaction (RT-PCR) positive nodes following sentinel node biopsy.[87,88] Finally, E1601 is aplanned study, which will compare 1 month of induction with HDIFN versus the full 1-year Kirkwood regimen in patients with microscopic lymph node involvement.[87]

3.1.2 Intermediate-Dosage Trials

EORTC 18952[58,85] tested IDIFN in AJCC stage IIB/III melanoma. This was administered during an induction phase (10 MU/day subcutaneously for 5 days per week over 4 weeks) followed by maintenance over 1 or 2 years (10 or 5MU three times weekly, respectively) and was compared with observation.[89] The first analysis of EORTC 18952 showed a benefit in distant metastasis-free interval (DMFI) for the 2-year IFNα arm (i.e. 5MU three times weekly). This benefit was statistically significant only for stage IIB disease.[89] Although IFNα at 5 to 10MU was well tolerated, the effect on OS was not significant. This may change with longer follow-up (so far, median 1.6 years).

The Scandinavian Melanoma Co-operative Group is conducting a similar trial with IDIFN as induction followed by maintenance with 10MU three times weekly for 1 or 2 years versus observation.[87] The German Dermatology Cooperative Oncology Group (DeCOG) initiated a trial of IDIFN versus LDIFN for stage II disease in 1997.[59] The absence of a direct comparison of IDIFN with HDIFN in this study may create problems in interpreting OS results if one of these IDIFN arms shows a survival advantage.[87]

3.1.3 Low-Dosage Trials

The first of these studies was WHO-16, which examined the role of LDIFN (3MU three times a week subcutaneously for 3 years) versus observation.[70,71] Neither the initial analysis of WHO-16 in 1994[70] nor a more recent analysis[90] showed any differences in DFS or OS between the LDIFN and observation groups. WHO-16 focused on recurrent nodal disease. It is possible that recurrent nodal disease is inherently resistant to IFNα, thus explaining no survival benefit when adjuvant IFNα is used in this setting. Alternatively, it is also possible that the lack of activity of IFNα may be related to the lower dosage used (approximately 40% of the total cumulative dose) compared with the Kirkwood regimen. The latter possibility, however, is difficult to reconcile in the light of two recent French and Austrian studies, which showed better DFS for LDIFN-treated patients (68 vs 58% and 72 vs 64%, respectively, at 3 years). In these two studies, IFNα was given for 12 to 18 months to patients with stage IIA-B melanoma, compared with 36 months in the WHO-16 study, where the patients had stage III disease (i.e. nodal metastases). These findings are difficult to explain. One could speculate that LDIFN (about 3MU three times weekly) given for long enough (>1 year) may be effective for just stage II disease. However, in the French study, the significant improvement in OS at 5 years has been lost with longer follow-up.[91] In contrast, IFNα may need to be given in much higher dosages during induction and maintenance therapy for patients with nodal metastases (stage III).

The UK Adjuvant Interferon in Melanoma High Risk (AIM High) study compared LDIFN given over 2 years and observation for stage IIB and III melanoma. Data presented at 2001’ s American Society of Clinical Oncology (ASCO) meeting failed to demonstrate a survival advantage in favour of LDIFN.[81] A smaller study from Scotland, which used IFNα 3MU three times weekly for just 6 months, also failed to show a survival benefit in stage II disease.[83] The lowest dosage of IFNα has been tested in the EORTC 18871 trial. This was a four-arm study in AJCC stage IIB/III disease. It compared LDIFN with IFNα, a control group and the homeopathic extract ‘Iscador’ given only within the DKG-80-1 German Cancer Society trial. LDIFN had no effect on DFS or OS.[89]

3.2 Metastatic Disease

3.2.1 Interferon-α Alone

Metastatic malignant melanoma has a poor survival of just 6 to 9 months.[92] Although dacarbazine is the most commonly used cytotoxic agent in metastatic malignant melanoma, responses are seen in just 15 to 25% of cases[69] and only a minority of these are complete or durable.[93] Investigators have therefore conducted numerous trials over the last 15 years in an attempt to improve response rates and survival. In the 1980s and early 1990s, many phase II studies (summarised in table IV) were conducted using IFNα in metastatic malignant melanoma.[60–68,94–99]

Table IV
figure Tab4

Phase II trials of single-agent interferon-α in metastatic melanoma

Full size image

These trials yielded response rates of about 15% (range 0 to 38%) with about 5% complete response rate.[101,102] Median duration of response was short (6 to 9 months) and at best 12 months.[103,104] The interferon dosages used were 3 to 50 MU/m2/day three times weekly, with 10 to 20 MU/m2 being the most common dosage. Uninterrupted (i.e. daily or thrice weekly) administration of interferon was more effective than interrupted (i.e. cyclic) treatment,[102] and responses were seen with dosages between 10 and 50 MU/m2/day on alternate days.[101] Moreover, tumour response was primarily limited to patients with small volume or soft tissue disease (including skin).[105] This and localised lung disease were among the predictive factors for response.[106] The main toxicities were flu-like symptoms (fever, chills, fatigue, myalgia and headache), nausea, leucopenia and neuropsychiatric disturbances (mood swings, cognitive dysfunction).[106]

These trials were important because they demonstrated that, although the response rate to IFNα was comparable to that with chemotherapy alone, the responses in a small fraction of patients were durable.[58,69] Although in recent studies of chemotherapy for advanced malignant melanoma survival beyond 3 years is very rare,[107] several phase II studies have reported long-term responses of 3 to 5 years using rIFNα alone.[65,94–96] To our knowledge, there are no data from randomised studies comparing rIFNα alone with chemotherapy in metastatic malignant melanoma.

3.2.2 Interferon-α with Interleukin-2 or Interferon-γ

IL-2 was combined with IFNα because IL-2 alone has shown response rates of approximately 23% in phase II trials[108] and synergy with IFNα has been demonstrated in vitro.[106] In phase II studies of IFNα + IL-2, response rates ranged from 0 to 44% (table V).[106] Initially, LDIFN + low-dosage IL-2 given subcutaneously showed little activity. In a dose-escalating study, Rosenberg and colleagues[109] reported response rates as high as 44%. Kruit and co-workers[110] later reproduced these findings. The optimal regimen was found to be IFNα 3 MU/m2 + IL-2 11.7 MU/m2 intravenously every 8 hours for 5 days. These and other phase II studies suggested the existence of a dose-response relationship with IFNα + IL-2.[106] Unfortunately, the severe toxicities associated with the higher dosages precluded routine clinical use. This led to the development of the ‘decrescendo’ schedule by Keilholz et al.[111] This used a rapidly tapering dosage of IL-2 with IFNα 10 MU/m2/day subcutaneously on days 1 to 5, 4-weekly) and achieved a response rate of 41%. Unfortunately, the only randomised trial, to our knowledge, of IFNα + IL-2 versus IL-2 alone was disappointing, with low response rates of 10 and 5%, respectively, and no survival difference.[112] IFNα has also been used with IFNγ and found to be ineffective.[113]

Table V
figure Tab5

Phase II trials of interferon (IFN)-α with interleukin (IL)-2 in metastatic melanoma

Full size image

3.2.3 Interferon-α with Single-Agent Chemotherapy

IFNα has been combined with cytotoxic chemotherapy, mainly dacarbazine (table VI) and cisplatin. Dacarbazine is the most tested cytotoxic agent in melanoma with one of the highest response rates (approximately 20%).[118] One small randomised trial of 61 patients showed a significant improvement in response rate and survival for the IFNα + dacarbazine combination.[119] Three more studies involving more than 650 patients failed to confirm the earlier findings.[120–122] Consequently, the combination of IFNα with dacarbazine cannot be recommended.[106]

Table VI
figure Tab6

Randomised trials of dacarbazine (DTIC) ± interferon (IFN)-α in metastatic melanoma

Full size image

Cisplatin is one of the other active agents in metastatic malignant melanoma, with response rates of 15 to 25%.[118] A small phase II study showed that IFNα + cisplatin was tolerated well and that some synergism may exist between them.[123] Despite encouraging preclinical data, phase I/II studies involving about 100 patients have reported response rates of 7 to 14%,[118] with no improvement in survival.[123,124]

3.2.4 Interferon-α with Combination Chemotherapy

Several phase II trials in which 317 patients have been treated have demonstrated improved response rates (26 to 68%, mean 44%) when IFNα is combined with several cytotoxic agents (table VII).[125–131] Pyrhonen and colleagues[125] reported one of the highest response rates; however, nearly half of their patients had stage III disease (i.e. nodal metastases), which carries a better prognosis than stage IV disease.[106] The combination of IFNα with dacarbazine and fluorouracil protected against central nervous system relapse in responding patients,[132] and this should be evaluated further. Nevertheless, none of these studies showed improvement in survival, and because of the enhanced toxicity seen with some regimens the use of IFNα with combination chemotherapy outside a clinical trial is not recommended.

Table VII
figure Tab7

Phase II trials of interferon-α with combination chemotherapy in metastatic melanoma (reproduced from Legha,[69] with permission)

Full size image

3.2.5 Interferon-α with Chemoimmunotherapy

IFNα has been combined with conventional cytotoxic agents with or without IL-2 and tested in more than 900 patients in phase III studies (table VIII).[125,132–137] Response rates have varied from 16 to 48% and median survival has been approximately 1 year. Over the last 10 years, some randomised studies[138–142] have compared IFNα-based immunotherapy with immunotherapy + chemotherapy. In three of six randomised trials of chemoimmunotherapy involving nearly 400 patients, there was a trend for a higher response rate (table VIII) in the IFNα or IFNα + IL-2 arms. However, none of these or any of the other studies demonstrated a survival advantage for either arm. In the 1997 study by Keilholz and co-workers,[139] addition of cisplatin to IFNα and IL-2 enhanced the response rate but did not affect survival, whereas in the study by Rosenberg and colleagues[141] (cisplatin, dacarbazine, tamoxifen ± IFNα and IL-2) there was no statistical difference in response rate or survival. More recently, the Dermatologic Cooperative Oncology Group reported on a randomised trail of IFNα with dacarbazine ± IL-2 and showed no difference in response rate or OS.[107]

Table VIII
figure Tab8

Randomised trials of chemoimmunotherapy [interferon (IFN)-α with interleukin (IL)-2 and combination chemotherapy] in metastatic melanoma

Full size image

The sequence of administration may be important. Platinum-containing chemotherapy given preferably before or concurrently with immunotherapy has produced the best results. In a phase II study of 60 patients (30 per group) where chemotherapy was followed by immunotherapy, the response rate was 73%; if immunotherapy preceded chemotherapy the response rate was 47%.[135] This is not surprising, as conventional cytotoxic agents reduce disease volume, thus facilitating the subsequent antitumour function of immune effector cells. For concurrent chemoimmunotherapy the response rate in 52 patients was 63%.[135] More recent data have shown the response rate of concurrent chemoimmunotherapy and sequential chemotherapy-immunotherapy to be 62 and 60%, respectively.[137]

rIFNα has been tested with a variety of other agents such as cimetidine,[143] eflornithine (difluoromethylornithine),[144] low-dose cyclophosphamide,[145] retinoids[146,147] and indomethacin.[148] None of these has shown a potential benefit. Temozolomide has been tried in a phase I study with some responses.[149]

4. Renal Cell Carcinoma

IFNα has been extensively investigated in both the metastatic and adjuvant setting of renal cell carcinoma (RCC). RCC has a unique and complex natural history and can remain unrecognised during the lifetime of the affected individual, even after metastatic spread.[150,151] Even when clinically manifest with metastasis, RCC can undergo rare but well documented spontaneous remissions.[152,153] These facts indicate the ability of the host immune system to control RCC under certain, unique but undefined, cellular conditions. This, along with the evidence of efficacy of active specific immunotherapy, gave impetus to the use of IFNα in the management of RCC.[154]

4.1 Metastatic Disease: Monotherapy

IFNα was found to be active against RCC as early as 1983.[155] In vitro studies have also shown that the interferons have a direct antiproliferative effect on renal tumour cells.[156,157] The response rates for metastatic RCC, in various phase II studies, range from 0 to 31% (table IX). Even though most of the responses are partial responses and only a few are durable complete responses, the responding patients were found to have significant prolongation of progression-free survival. These results led to the evaluation of IFNα in randomised phase III trials, which have confirmed the efficacy of IFNα in metastatic RCC. However, the response rates in the randomised studies are only in the region of 10 to 15%.[158] One of the early randomised trials using an IFNα dose of 2MU found no benefit for IFNα compared with medroxyprogesterone.[159] The negative result was probably due to the low dose of IFNα used in the trial, since a previous phase II study and a randomised trial had shown lack of benefit for LDIFN compared with HDIFN.[160,161] Another randomised trial in the 1990s also failed to confirm the beneficial effects of IFNα.[162]

Table IX
figure Tab9

Interferon (IFN)-α monotherapy in metastatic renal cell carcinoma (survival data for randomised studies only)

Full size image

The first major randomised trial to demonstrate a survival benefit for IFNα monotherapy in metastatic RCC was the Medical Research Council (MRC) RE 01 trial. This trial randomised patients with advanced RCC to either medroxyprogesterone or IFNα.[170] A total of 350 patients were randomly assigned to receive treatment, and by the time early results were published 236 patients had died of disease, with 90% dying of metastatic RCC. There was an OS advantage for IFNα, with a hazard ratio of 0.72 (0.55 to 0.94). The absolute improvement in 1-year survival for the interferon group was 12% (43 vs 31%). The improvement in median survival of 2.5 months due to IFNα has to be viewed against significant toxicity, namely tiredness, nausea and lack of energy, which may impair quality of life. The recently updated results have confirmed the results of earlier analysis.[171]

The modest efficacy of IFNα in metastatic RCC has to be balanced against treatment-related toxicity. An attempt is being made to improve the therapeutic ratio of interferon by modifying its structure. One such modified molecule is pegylated interferon (peginterferon-α), which after subcutaneous administration results in sustained absorption. Peginterferon-α is IFNα modified by the covalent attachment of polyethylene glycol polymer molecules; this renders it less immunogenic and prolongs its half-life.[172] A recent phase II trial of peginterferon-α in 40 patients has shown an response rate of 13% and favourable toxicity profile, and hence phase III trials are planned.[173]

4.2 Metastatic Disease: Combination Therapy

Even though IFNα produces a small but significant survival advantage in the treatment of metastatic RCC, the response rates have been fairly modest (<20%), particularly in the randomised trials. This has led to attempts to integrate IFNα with other biological agents and conventional chemotherapy agents. The widely evaluated combinations are IFNα + IL-2, and the addition of cytotoxic agents to IFNα (table X).

Table X
figure Tab10

Interferon (IFN)-α combination therapy in metastatic renal cell carcinoma (survival data for randomised studies only)

Full size image

4.2.1 Interferon-α with Interleukin-2

IL-2 is predominantly used for metastatic RCC in North America, whereas IFNα is used preferentially in Europe. Both have only modest efficacy, and hence attempts are being made to give them in combination, with a view to improving the outcome of RCC. The combination of high-dosage IL-2 intravenously and IFNα subcutaneously produces significant toxicity.[168,209] Hence, alternative strategies were explored to reduce toxicity. One such approach was subcutaneous administration of low-dosage IL-2, which is associated with less toxicity.[175,176,210] Even so, the combination of subcutaneous IL-2 and IFNα has also been found to have significant toxicity in a few studies. Moreover, the low dosages of the cytokines could compromise efficacy of the regimen.[182,211,212]

One early, multicentre, phase II study assessed the feasibility of combining IL-2 with IFNα using an outpatient low-dosage subcutaneous regimen. This study involved 42 patients with immunotherapy-naive metastatic RCC and found an response rate of 12% with mild toxicity and a median survival of 14.5 months.[176] The response rates of up to 30% in various pilot studies led to prospective randomised studies of the IFNα + IL-2 combination.[174] The results from the prospective randomised studies did not confirm the results of the pilot studies.[178–182]

A multicentre randomised French trial[183] provides interesting data regarding the IFNα + IL-2 combination. In this trial, 420 patients were randomly assigned to continuous intravenous IL-2 or IFNα, or both. After a median follow-up of 39 months, the OS was not significantly different between the three groups even though the response rate in the combined arm was more than twice the response rate in the single-drug arms. The response rates in the single arms involving IL-2 and IFNα were low, at about 7 and 8%, respectively. Moreover, toxicity was higher in the IL-2 arms. This trial effectively rules out the use of this combination in the routine clinical setting as well as in the trial setting. Moreover, the toxicity profile in the three arms of this trial favours the use of IFNα over IL-2 in the routine clinical setting.[183]

The combination of low-dosage IL-2 and IFNα has been compared with high-dosage IL-2 in a recent phase III randomised trial.[184] A total of 193 patients were randomised in this trial and the preliminary results, in an abstract, show favourable response rates in the high-dosage IL-2 arm (25 vs 12%), thereby suggesting that the combination of subcutaneous low-dosage IL-2 and IFNα is unlikely to produce superior survival benefits.[184]

Attempts are also being made to improve the therapeutic ratio of the IFNα + IL-2 combination by adding a chemotherapeutic agent. The drawbacks of this approach are that the increased toxicity compromises the cytokine dosage and the regimen affects the quality of life of patients who essentially have incurable disease. A phase II study assessing the benefit of adding vinblastine to the combination of IFNα + IL-2 showed a response rate of 38%. About 50% of patients experienced grade 2 to 3 neutropenia, but this did not warrant dosage modification. These results need further investigation in a randomised trial.[198]

A few trials have examined the value of adding isotretinoin (13-cis-retinoic acid) to the combination of IFNα and IL-2, but the value of this approach remains uncertain.[206,213]

4.2.2 Interferon-α with Interleukin-2 and Fluorouracil

The combination of fluorouracil, IFNα and IL-2 was pioneered by Atzpodien’s group.[185] This outpatient combination therapy in a small trial of 35 patients produced an response rate of 48% (complete response 11% + partial response 37%), with a further 37% having stable disease. Moreover, the systemic toxicity in this trial was only mild to moderate. Although many other groups have noted similar results for this combination,[186,188,189,191,193] recent studies, including a randomised trial, have not confirmed its superior response rate and better tolerability.[177,187,190,192,194–197,214,215] The randomised phase III MRC RE-04 trial, which is comparing the three-drug ‘Atzpodien’ regimen with single-agent IFNα, will clarify whether this combination is superior to monotherapy in metastatic RCC. The combination of fluorouracil + IFNα + IL-2 is also being tested in an EORTC adjuvant trial of high-risk RCC patients.

4.2.3 Interferon-α with Vinblastine

The preclinical data suggesting synergism between IFNα and vinblastine led to pilot clinical studies that showed response rates of 16 to 44%.[216–218] Conversely, an early prospective study and two randomised studies failed to demonstrate any significant benefit for the combination.[199,219,220] By contrast, the recently published trial by Pyrhonen and colleagues[200] was the first trial to demonstrate survival benefit for combination therapy in metastatic RCC. This prospective randomised trial demonstrated that the combination of IFNα, at a dosage of 18MU three times weekly, with vinblastine was superior to the control arm of vinblastine alone. This trial randomised 160 patients and the median survival of patients receiving combination therapy was 67 weeks as opposed to 37 weeks in the control arm. Unlike this trial, an earlier European multicentre study, using IFNα as the control, found no survival benefit for the combination of IFNα and vinblastine, even though there was a doubling of the response rate in the combination arm.[199]

4.2.4 Interferon-α with Retinoids

Preclinical studies demonstrating synergism between IFNα and the retinoids[221] led to phase II trials in which response rates of up to 30% were noted for this combination.[222] However, a recent randomised study has not shown any significant benefits in terms of response rate or OS for the addition of isotretinoin to IFNα.[223] A positive aspect of this trial was that the duration of response and progression-free survival were longer in the combination arm, but this came at the price of poor quality of life.[223] A more recent study combined LDIFN with isotretinoin and weekly paclitaxel chemotherapy in a group of patients with advanced RCC, some of whom were pretreated with IL-2 and/or IFNα. Although this treatment was well tolerated, its efficacy was minimal.[224]

4.2.5 Interferon-α with Interferon-γ

Once again, preclinical studies suggesting synergism between IFNα and IFNγ led to clinical trials evaluating this combination.[225] An EORTC phase III trial exploring this combination failed to show beneficial effects.[202] An earlier small randomised study by Foon and coworkers[226] again showed a lack of benefit for the combination. Together, these studies confirm the lack of synergism between IFNα and IFNγ in metastatic RCC.

4.2.6 Interferon-α with Fluorouracil

The combination of IFNα and fluorouracil showed response rates of 43% in one of the early studies,[205] but many other studies suggest that this combination is unlikely to produce a better outcome in patients with metastatic RCC.[227,228]

4.2.7 Other Combinations

IFNα has been combined with many other agents in various phase I/II trials with mixed and inconclusive results. Randomised studies exploring the addition of cimetidine + coumarin, or aspirin (acetylsalicylic acid), to IFNα found no difference in response rate or OS.[203,204] Histamine protects immune cells (NK and T cells) by inhibiting the formation and release of reactive oxygen species, and hence it is being assessed in combination with IFNα and IL-2. The phase II results are interesting and further trials are needed before firm conclusions can be drawn.[208]

4.3 Dose Response, Treatment Duration and Late Responses

The dose of IFNα used in the various studies of RCC ranged from 1 to 30MU, typically given three times a week by subcutaneous route. The lower doses have been found to be ineffective in phase II and phase III trials.[160,161,167] The efficacy of IFNα reaches a plateau after doses of about 10MU, and higher doses are associated with inferior response rates because of poor tolerability.[168] A bell-shaped curve has been suggested for the dose-response curve.[229] Hence, IFNα doses of the order of 5 to 10MU are commonly used in the clinical setting.

The optimal duration of IFNα treatment is yet to be well defined. The MRC RE 01 randomised trial, which was the first to demonstrate a survival benefit for IFNα therapy, used an IFNα schedule of 10MU three times weekly for 12 weeks only. But the evidence indicates that prolonged IFNα therapy would be beneficial. Recently, in a phase II trial of 75 patients, Kankuri and colleagues[230] observed an response rate of 17% (43% stable disease) with the use of an intermittent schedule of IFNα therapy [treatment with maximum tolerated dose (MTD) with escalating doses of 4.5 to 18MU, three times weekly, with 1 week rest each month]. Significantly, they observed three late objective responses (4%) after a long period of stable disease (12 to 16 months). The responding patients were noted to have prolonged median time to response (median 7 months; range 4 to 16 months). Despite the monthly interruption of treatment for a week, there was no apparent adverse effect on the response rates and survival rates. The median duration of response was 16 months, and indirect comparisons indicate that the survival rates are better than those previously reported with continuous IFNα regimens.[231,232] Moreover, the objective responses were noted only in patients treated with 9MU, and 88% of the patients tolerated an IFNα dosage of at least 9MU.[230] This trial demonstrates the need for prolonged IFNα therapy (>12 months) at MTD, not only in patients with objective response but also in those with stable disease.

Buzio and coworkers[233] provide further support for prolonged IFNα therapy in RCC. In a phase II trial of 50 patients, they administered low-dosage IL-2 (1 MU/m2, days 1 to 5) and IFNα (1.8 MU/m2, days 3 and 5) for 4 consecutive weeks every 4 months, irrespective of response. The overall response rate was 12%. They observed four late responses (8%) after a period of 14 to 16 months in patients who were treated with immunotherapy despite progressive disease. Hence, IFNα therapy should be administered indefinitely to patients with responding or stable disease. The role of low-dosage intermittent immunotherapy needs further evaluation in large randomised studies.

4.4 Adjuvant Therapy

About 15 to 50% of patients with localised RCC will eventually relapse. In particular, a majority of patients with resectable but locally advanced node-positive tumours will relapse eventually.[234–237] Since IFNα has modest efficacy in metastatic RCC, studies were initiated to assess the effect of IFNα in the adjuvant setting, where it was hoped to be more effective because of the low tumour burden. The early pilot studies showed a significant increase in host NK cell activity after initiation of IFNα therapy. This immune modulation was expected to reduce the risk of recurrence in high-risk patients after nephrectomy.[238] However, randomised studies of IFNα in the adjuvant setting have not shown any survival benefits or reduction in recurrence rates.[239–242]

5. Summary and Conclusions

5.1 Malignant Melanoma

A recent meta-analysis of the four ECOG trials of HDIFN[243] has reported a beneficial effect of HDIFN on DFS and OS. One must, however, take into account the fact that the median follow-up time for E2696 and E1694, which together account for nearly half of all patients, was relatively short (22 and 29 months, respectively). More importantly, at median follow-up of 12 years, the OS benefit for the HDIFN arm in E1684 has been lost.[78] When only data from mature HDIFN trials are considered, it appears that HDIFN prolongs DFS[78] (supported by the E1684,E1690 and E1694 data) but not OS (supported by E1690 and the latest data from E1684). The recent results of E1694 could provide support for HDIFN; however, in addition to the above, two important issues remain to be resolved. First, the E1694 data must be allowed to mature and, secondly, reproducibility of the E1694 findings is essential before a change in clinical practice occurs. Moreover, the toxicity and financial costs associated with HDIFN are so significant that there must be unequivocal evidence for the benefit of such treatment before it is universally applied.

A meta-analysis done by the UKCCCR includes data from trials comparing adjuvant IFN with observation (about 3000 patients in total). The meta-analysis, which included data from EORTC 18871, the AIM High study and the Scottish study, showed a statistically significant benefit in DFS (p = 0.00001) and OS (p = 0.02).[58] A more recent meta-analysis[82] that included data from ten randomised trials (about 3700 patients), and included the latest AIM High results,[81] has again shown a benefit in DFS in favour of IFNα [odds ratio (OR) 0.84, 95% confidence interval (CI) 0.77 to 0.92, p = 0.0001]. For OS, however, the benefit was less clear (OR 0.90, 95% CI 0.82 to 1.00, p = 0.05). Interestingly, there was no statistically significant evidence of benefit of HDIFN versus LDIFN. As far as IDIFN is concerned, no conclusions can be reached until the results of EORTC 18952 mature. The ongoing EORTC 18991 is a randomised study evaluating long-term administration of peginterferon-α (administered once weekly for 5 years) in stage III disease compared with observation.[91] LDIFN has not been shown in any study to prolong OS, and therefore its routine clinical use outside a clinical trial cannot be recommended. Others have echoed this view even before the publication of the AIM High study.[58,106,244] A more recent meta-analysis[245] of eight randomised, controlled clinical trials of adjuvant LDIFN or HDIFN in which 3178 patients were treated has shown no clear benefit in OS. Some investigators believe that adjuvant HDIFN should be considered standard therapy for high-risk patients who are likely to tolerate it.[246] However, the absence of a convincing benefit on OS as demonstrated by the two recent meta-analyses[82,245] would not be supportive of the use of adjuvant IFN outside a clinical trial.

In advanced melanoma, the in vitro data that showed synergism between IL-2 and IFNα[31,247] have not translated into a clear-cut survival benefit. On current evidence the IFNα + IL-2 combination cannot be recommended for routine clinical use. What has emerged, however, is that a small percentage of patients treated with IFNα + IL-2 will have durable responses.[69,108] In one study, nearly 40% of patients survived for more than 3 years.[248] One of the challenges for the next few years, therefore, will be to establish tumour and patient characteristics that predict long-term responders.

Chemoimmunotherapy (‘bio-chemotherapy’) has achieved high response rates[69,106,108] (40 to 55%), and high complete response and long-term non-maintained remission rates (5 to 15%).[118] An analysis of trials that included regimens with IFNα + IL-2 showed a 2.5-month survival advantage compared with chemotherapy alone.[249] However, a more recent retrospective analysis of 15 phase II/III trials suggested that IFNα had no effect on OS.[250] Chemoimmunotherapy regimens may increase toxicity significantly,[251] and as their benefits are unproven[8,106] routine use outside prospective randomised trials cannot be recommended.[251] The currently ongoing study ECOG 3695/CALGB 509802 by the Eastern Cooperative Oncology Group and the Cancer and Leukaemia Group B comparing cisplatin, vinblastine and dacarbazine (CVD) chemotherapy with CVD + IL-2 + IFNα may give some answers.[8] Although the role played by IFNα in the induction and maintenance of remission with bio-chemotherapy is still unknown, the above information should be taken into consideration in the design of future phase III randomised trials.

5.2 Renal Cell Carcinoma

IFNα has definite but modest activity in metastatic RCC. Significant long-term survival occurs in a small cohort of patients as a result of IFNα treatment.[252] This has to be balanced against the toxicity of the treatment, the impairment in quality of life of many to benefit a few and, importantly, the high cost of IFNα therapy. Hence, proper selection of patients for treatment is very important. The response to IFNα is associated with various prognostic factors such as good performance status, solitary or fewer metastatic sites, prior nephrectomy and low erythrocyte sedimentation rate.[169,253,254] These factors should guide the initiation of IFNα treatment in metastatic RCC. Of special importance is the issue of prior nephrectomy. Two randomised studies[255,256] demonstrated that nephrectomy prior to immunotherapy with IDIFN (5 MU/m2 subcutaneously three times weekly), in patients with advanced RCC confers a survival benefit and delays time to progression. Hence, nephrectomy should be considered at least in patients with good performance status (WHO 0 or 1).[257]

In the adjuvant setting, there is no evidence for use of IFNα outside clinical trial protocols. Similarly, the evidence is also weak for use of IFNα in combination with IL-2 and other chemotherapy agents. Peginterferon-α may have the potential to improve the therapeutic ratio of IFNα treatment, and this needs assessment in large randomised trials.[173]

The second part of this review[258] will deal with the use of IFNα in solid tumours where it has demonstrated activity, such as superficial bladder cancer and Kaposi’ s sarcoma. Part 2 will also cover head and neck cancer, gastrointestinal malignancies, common solid tumours such as lung and breast cancer, and gynaecological malignancies.