1. Clinical trials
  2. Adverse Effects
  3. Predicting and monitoring treatment - biomarkers
  4. Some specific immunotherapy aspects of head & neck malignancies

Immunotherapy is a rapidly evolving area in the treatment of many malignancies, with therapies based on checkpoint inhibitors and vaccination concepts making particularly good progress. The importance and potential of immunotherapies for the future treatment of many malignancies has been recognised not least by the award of the Nobel prize for physiology or medicine in 2018 to J.P. Allison (immune checkpoint therapies targeting CTLA protein) and T. Honjo (targeting the PD-1 protein and receptors) for their ground-breaking research into the development of such therapies.

Immunotherapies have had spectacular successes and failures so far. Some have recently become more widely used in clinical studies; some have become well established in the treatment of, for example, melanoma; some are in the pre-clinical stages of testing. There is clearly an urgent need to learn and understand much more about the intricate regulation and interaction networks of the immune system itself, in sickness and in health, and its interactions with the various currently available immuno-modulating drugs.

At this early stage of clinical development of immunotherapies, it seems appropriate to look a little closer into the processes of fundamental changes and new drugs entering the mix of clinical treatment modalities. We look into the (minefield of) basic aspects of clinical trials. Adverse effects (as far as known to date) and the lack of suitable biomarkers to predict and monitor treatment efficacy and progress need to be considered. Finally, some aspects specific to head & neck malignancies in relation to immunotherapies are discussed. We are well aware that, given the rapid developments in biochemistry and microbiology, this page may have to be updated or re-written regularly and/or may have to be moved to the section on this website dealing with established treatment modalities in the not too distant future.

Clinical trials

Clinical trials are established procedures to test a treatment modality, old or new, for a disease or condition, to help decide whether or not it should be used to treat the general population in the future (or continue to be used as practised in the past). There are a number of general ethical and technical issues arising from this approach (see below).

In the assessment of new treatment modalities, there will have been extensive testing done in vitro (on microorganisms or isolated cells in the laboratory), followed by testing on animals before clinical trials involving humans are carried out. This staged approach is thought to minimise the risk to participants in human clinical trials, but not all risks for humans can be excluded in this way. Many clinical trials are randomised, with participants picked at random and assigned to (often) one of two groups, receiving a particular treatment or a placebo, with the participant not knowing to which group they were assigned. Many clinical trials, where this is technically possible, are ‘double blind’, meaning that neither the participant nor the health care team know to which group the participant has been assigned. This approach is thought to avoid bias. It has to be said, however, that an uncompromising determination to exclude everything that may be considered biased, may carry additional risks for participants.

Some aspects about design and execution of clinical trials

Clinical trials are attempts to evaluate the efficacy and other aspects of old and existing treatment modalities in an unbiased manner. The approach relies heavily on numerical and statistical methods. There are two fundamental aspects to this. First, any statistical evaluation, in clinical trials or otherwise, can only ever be as good as the data / observation input available. Second, it is crucial to keep in mind that even if the statistical output / results are definitive-looking numbers, the very nature of such statistical analyses is probabilistic. One may think of it as some form of numerically informed interpretation of the observations of particular treatment modalities. The flavour of statistical data analysis in clinical trials (or, say social sciences and similar) is thus quite different from the statistical analysis of uncertainties, variance and other error evaluations of an array of measurements in a physics laboratory. From this it follows directly that not all clinical treatment modalities will be suitable for ‘quantification’ by meaningful statistical analysis. Also, where a clinical-trial approach holds promise for a meaningful and informative statistical result in principle, in practice careful design and execution of a trial is essential to obtain any meaningful results.

Planning and execution of clinical trials has two major components, closely related to each other. One aspect concerns the size of the cohort of participants required for statistically (and clinically) meaningful results. This requirement of the size of the group is mainly dictated by the expected results, if a very clear distinction in a homogeneous group of participants is to be expected, or if a trial is looking for more subtle effects. In a nutshell, if major effects and differences are expected, a smaller number of participants (divided into sub-groups, receiving treatment or placebo) is needed, whereas a significant (statistically and otherwise) result when studying more subtle effects, possibly in a more heterogeneous group of participants, requires a larger cohort. Any clinical trial has to address these aspects in the planning as well as in the execution stages.

Whilst by design meant to exclude, or at least minimise, bias in clinical trials, the very nature of how such a trial is set up, is a source of bias and error in its own right. What should be the inclusion and exclusion criteria for participants for a particular trial? What should be the measure(s) / criteria to define success or failure of a treatment modality? For how long does a trial need to run? Can / should the trial be double blinded, or not? Does the trial need to be multi- or single-centred? What will qualify as a threshold for terminating a trial early? Many more questions come to mind, and the somewhat sobering conclusion at this stage has to be the following. The design and the execution of clinical trials can fall foul to genuine human error (there is a tragic such example of otherwise well executed clinical trials about adjunct radiotherapy of oropharyngeal cancers after surgery. These were carried out at a time when it was not well enough understood that HPV-positive oropharyngeal squamous cell cancers are a different disease from HPV-negative ones. In consequence, many patients with HPV-positive malignancies were overtreated, with lasting serious quality-of-life issues in many cases). The design and the execution of clinical trials on the other hand also lends itself to gaming the system. For example, a suitably and narrowly defined group of participants may demonstrate benefits of a particular drug or other treatment modality, but may not at all be representative for a corresponding ‘real world’ group of patients.

We certainly do not advocate that clinical trials should be abolished, but we wish to emphasise the importance of technical expertise in statistics, the need to conduct trials in a responsible manner, and not to be fooled to think that just because trials ‘produce numbers’ that the meaning and clinical significance are in any way simple black-and-white affairs. A critical-thinking approach to clinical trials appears more important than ever, given that ‘the randomised clinical trial’ has acquired some kind of gold-standard status in the medical world. A study is not necessarily a good one, just because it happens to be technically a randomised clinical trial. Also, it has to be recognised that other types of study and assessments in medicine do have their place and in some instances are more appropriate than clinical trials. Not everything can or should be measured or attempted to be expressed in a single number. Given the widespread use of clinical trials and all the underlying difficulties and problems, it is not surprising that there are ongoing debates about best practice with regard to statistics and design & execution of clinical trials.

Ethical considerations about trials

It is important to understand that the purpose of a clinical trial is not primarily the treatment of the participants, but to gain medical knowledge for future improved treatment of patients. That mission and the means to achieve it in itself carries considerable ethical conflicts. Obviously, informed consent by all participants to take part in a trial is a prerequisite. This sounds obvious and straightforward in theory, but is difficult to achieve and to ascertain in reality. It certainly requires that all participants understand not only the purpose of the trial, but also all known risks and benefits of the trialled treatment.

In trials with randomised groups, half of the participants receive a placebo. The use of placebos in general is debated: participants may be led to believe that they receive a particular drug when they are not; receiving a placebo instead of the trialled drug may be an advantage or a disadvantage. For example, if the studied treatment turns out beneficial, the participants in the placebo group will not have received the best standard of care. In either case, the use of placebos carries a certain element of deceit even though the ’placebo effect’ is well recognised in, for example, pain studies. Randomisation raises similar ethical issues. Double-blinded studies have an additional burden of ethical issues for participants and clinicians. There is potentially an enhanced risk for the participants associated with the clinicians not knowing who is receiving placebo or active drug.

Technical and statistical considerations at the planning and implementation stages of a clinical trial seamlessly lead to, and overlap with, ethical considerations. It is compulsory to keep any risks for participants at an absolute minimum, and one of the best ways of assuring that, is to keep the size of the trial group at the smallest reasonable numbers of participants, so that nobody is needlessly exposed to risks. Managing and minimising risk to participants may also mandate stopping a trial earlier than anticipated. This may be necessary due to unexpected adverse effects, where the risks of continuing the trial would outweigh any potential benefits. If a study hypothesis is either proven early, or is shown to be unprovable or wrong, it would be unethical to continue a trial beyond that stage.

Another theme where technical and ethical considerations meet is the choice of measure(s) of success / failure. It is often said that concentration on metrics is biased (sic!) in favour of whatever is easy to enumerate. That, however, may not be the only and/or the most relevant measure. For example, overall survival time and progression-free survival period are the only measure used in the many clinical trials on immunotherapies, other considerations such as adverse effects or quality-of-life aspects are usually reported but not as a prime consideration in judging success or failure. This is a typical example where a ‘number’ only tells a part of the overall story and where clinical relevance should be more in the foreground: what is the point in having a marvellously effective drug if in the end it is too toxic for the intended clinical applications (bleomycin, originally designed as an antibacterial agent, turned out to be too toxic for this purpose but later found a niche application in chemotherapy and sclerotherapy due to its cytotoxic properties).  

Table 1 (checkpoint inhibitors) and Table 2 (monoclonal antibodies) below list some recent clinical trials in immunotherapy and their primary results, illustrating the current general picture and including another ethical dilemma: all trials were funded by the respective drug manufacturers. The two tables further document that clinical trials for head & neck malignancies are difficult to arrange when larger numbers of participants are needed, in line with other less common conditions. One may also read from the tables some bias in the reporting of adverse effects (see below) where the most common adverse effects are listed, but these often are not the most severe, or even fatal, ones (see below). It has to be said, though, that the practice with reporting adverse effects of immunotherapies is now changing. The slow take-up of systematic and full reporting of adverse effects of immunotherapies may also reflect some human-factor effects of denial, to some degree: it is a disappointment to the medical community if an effective medicinal drug turns out to be associated with severe adverse effects.

Table 1: Selected clinical trials – checkpoint inhibitors
Drug Trial funded by Summary of trial Adverse effects
pembrolizumab manufacturer of drug 132 participants with HNSCC (head and neck squamous cell carcinoma), adequate organ function, PD-L1 positive or negative. Treated for 2 years, until cancer progression, or a complete response, or withdrawal. Only 26 completed treatments. 6 months progression free survival 23 %, 6 month overall survival 59 %, median overall survival 8 months. Overall response rate higher in PD-L1 positive (22 %) and just 4 % in PD-L1 negative. in 82 participants; most common fatigue, hypothyroidism, and decreased appetite 15 withdrawals due to adverse effects no treatment related deaths
pembrolizumab manufacturer of drug randomised trial for non-small cell lung cancer after chemotherapy, with metastasis and some malignant cells producing PD-L1. 1034 participants three groups: 345 usual dose of pembrolizumab, 346 higher dose of pembrolizumab, 343 had docetaxel (chemotherapy). Overall survival 10.4 months for usual dose, 12.7 months for higher dose, 8.5 months for docetaxel. If high positive for PD-L1, pembrolizumab more effective for overall survival but no difference in progression-free survival. most common decreased appetite, fatigue, nausea, and hypothyroidism fewer adverse effects caused by pembrolizumab than by docetaxel 6 pembrolizumab-treatment related deaths
chemotherapy plus ipilimumab manufacturer of drug randomised trial of 502 participants with stage 3 or 4 advanced melanoma, with no previous treatment in two groups: 250 chemotherapy (dacarbazine) plus ipilimumab, 252 only chemotherapy. Only 92 participants completed four treatment cycles of combined therapy, 165 for chemotherapy only. Most commonly cancer worsening and serious ipilimumab-related adverse effects. Average overall survival just over 11 months (combination therapy) or 9 months (chemotherapy only). Survival after 3 years was 21 % (combination therapy) and 12 % (chemotherapy only). Survival after 5 years was 18 % and 9 %. more cases of severe adverse effects in the ipilimumab plus chemotherapy group than in the chemotherapy only group (56.3 % vs 27.5 %) most common immune system related adverse events such as elevated liver function, and diarrhoea
comparison ipilimumab and pembrolizumab   manufacturer of drug 834 participants with advanced melanoma; 278 ipilimumab and 279 pembrolizumab every 2 weeks, 277 pembrolizumab every 3 weeks. Treatment abandoned if the cancer progressed. 6 months progression-free survival 26.5 % for the ipilimumab group, 47.3 % for the pembrolizumab every 2 weeks, and 46.4 % for the pembrolizumab every 3 weeks groups. 12 months the overall survival 58.2 % for the ipilimumab group, 74.1 % for the pembrolizumab every 2 weeks group, and 68.4 % for pembrolizumab every 3 weeks group. pembrolizumab adverse effects less severe than for ipilimumab fewer participants terminating treatment early for pembrolizumab
nivolumab   manufacturer of drug randomised trial with 361 participants with head & neck squamous cell carcinoma, with progression or recurrence within 6 months of (platinum) chemotherapy; two groups: 240 nivolumab, 121 standard treatment (docetaxel or cetuximab). Median overall survival 7.5 months (nivolumab) and 5.1 months for standard treatment; both groups median progression free survival of around 2 months. less severe adverse effects in the nivolumab group than in the chemotherapy group common adverse effects in the nivolumab group fatigue, nausea, and skin rash 2 nivolumab-treatment related deaths
nivolumab   manufacturer of drug randomised trial with 272 participants with non-small cell lung cancer with progression or metastasis; in two groups: half nivolumab, half docetaxel (chemotherapy). Overall survival 9.2 months for nivolumab group, 6 months for docetaxel group. less severe adverse effects in the nivolumab group than in the chemotherapy group common adverse effects in the nivolumab group fatigue, decreased appetite, and physical weakness
Table 2: Selected clinical trials – monoclonal antibodies
Drug Trial funded by Summary of trial Adverse effects
chemotherapy (irinotecan) and cetuximab manufacturer of drug 329 participants with bowel cancer, after chemotherapy and progressing; two groups: 219 combination treatment,111 had only cetuximab. Overall response rate 22.9 % with combination treatment,10.8 % with cetuximab only. adverse effects were more common in the combination-therapy group (65.1 % vs 43.5 %) most common adverse effects diarrhoea, physical weakness, acne-like rash
chemotherapy and cetuximab manufacturer of drug 130 participants with advanced bowel cancer and the normal K-RAS gene (as previous trial demonstrated that cetuximab does not work with mutated K-RAS gene). All received 12 weeks of chemotherapy and cetuximab. Then 78 had intermittent chemotherapy and intermittent cetuximab, and 91 had intermittent chemotherapy and continuous cetuximab. Median progression free survival was 12.2 months in the intermittent group, and 14.3 months in the continuous group, with median overall survival being 16 months, and 17.5 months. adverse effects were similar in each group most common effects skin rash, diarrhoea, and physical weakness   3 treatment-related deaths
chemo-radiotherapy and cetuximab manufacturer of drug randomised trial with 258 participants with non-spread oesophageal cancer. All given chemotherapy and 5 weeks of radiotherapy, half were also given cetuximab. Progression free survival at 24 weeks was 66.4 % with cetuximab and 76.9 % without cetuximab. Median overall survival was 22.1 months with cetuximab and 25.4 months without cetuximab. more cases of adverse effects in the cetuximab group most common effects skin problems, fatigue and dysphagia 3 treatment-related deaths

Adverse Effects

After decades of use, the serious short- and long-term adverse effects of radiotherapy and chemotherapy in the treatment of malignancies are well documented and reasonably well understood (with some notable exceptions). Both radio- and chemotherapy have been denoted as blunt agents that do not sufficiently discriminate between healthy and diseased cells and thus also damage healthy tissues. One of the hopes associated with the development of immunotherapies is a lower burden of adverse effects. It is too early for a comprehensive picture of any long-term adverse effects the currently used immunotherapy drugs may have. Patterns of acute and medium-term adverse effects are now emerging, and are increasingly reported in the literature.

Immunotherapies are systemic treatments and thus share some aspects of ‘bluntness’ with the traditional chemotherapy drugs. Traditional chemotherapy drugs are cytotoxic agents that do harm predominantly to cell lines with fast-growth characteristics, such as malignant tumours but also have cytotoxic effects elsewhere in the body, especially on tissues with a healthy rapid turnover, such as the oral mucosa. The working principles of immunotherapy agents are fundamentally different from those of traditional chemotherapy agents, and instead rely on interfering with the body’s immune system and trying to activate these body-defence mechanisms against malignant cells. The immune system operates all over the body. Accordingly, adverse effects related to the immune system can occur everywhere in the body. It is perhaps telling that recently the range of adverse effects of immunotherapies has been given a name, immune related adverse effects, IREAs.

Immune related adverse effects are common. IREAs affect up to 90 % of people treated with immune checkpoint inhibitors and occur in 70 % of monoclonal antibody treatments. Most of the less severe effects concern skin and gut, the more severe IREAs typically involve the digestive tract and / or the endocrine system. Some IREAs are irreversible. A period of 3 to 6 months for the occurrence of adverse effects is typically reported, but it is not yet possible to rule out delayed effects. The risk of adverse effects seems to be correlated with dose. A well-known pattern from traditional radio- and chemotherapy treatment schemes emerges also for immunotherapies: multi-agent and/or multi-modality treatment schemes carry an increased risk for more severe adverse effects.

The symptoms of many IREAs resemble those of autoimmune conditions, such as arthritis or Sjögren’s syndrome. However, in most cases none of the antibodies identified in specific blood tests for autoimmune conditions are observed. It is not yet clear if pre-existing autoimmune conditions are particularly negatively impacted by immunotherapies. This is of practical importance because many autoimmune conditions, including various forms arthritis and type I diabetes, are common conditions.

Current picture of immune related adverse effects

The most commonly reported adverse effect is fatigue, which is difficult to assess and to relate to its exact causes. Fatigue may be a sign of tumour progression, it may occur as a secondary effect of some other IREA (for example, effects on the endocrine system, see below), or it may even be some specific form of chronic fatigue syndrome.

IREAs afflict various organs and tissues, including

Management of immune related adverse effects

To date, no effective methods to prevent or predict adverse effects of immunotherapies have been described. Some of the more severe adverse effects (see above) mandate extensive diagnostic explorations.

Any management of the symptoms of IRAEs heavily relies on steroids for their anti-inflammatory and general immunosuppressive properties. These can be administered topically or systemically, with some less severe IRAEs within the remit of local steroid applications. Some more severe IRAEs are refractory to systemic steroids and may require treatment with a TNF (tumour necrosis factor) inhibitor or other potent immunosuppressive agents to curb the inflammatory response. Some serious adverse effects may require termination of immunotherapy. Any deficiencies resulting from dysfunction of endocrine glands require replacement therapies for whatever hormones are insufficiently produced, alongside careful monitoring.  

Much of the use of anti-inflammatory and immunosuppressive agents in managing IRAEs is informed by their established use in treating autoimmune conditions. However, IRAEs and general autoimmune conditions are not identical, even if some symptoms may be very similar. The use of immunosuppressive agents to mitigate the adverse effects of (immune-activating) immunotherapies presents a dilemma. It is currently not clear if, how, and to what extent immunosuppressive agents negatively interfere with ongoing immunotherapies and/or what effects interrupted immunotherapy schemes might have.

Some immunotherapy treatments are already showing very good efficacy in treating some malignancies, for example in the treatment of melanoma. It is, therefore easy to predict that growing numbers of people will be treated with this type of drug, will survive for longer or will be cured and thus will present with lasting and/or late adverse effects of these treatment schemes (somewhat reminiscent of radiotherapy and late osteoradionecrosis). As some of these effects are difficult to diagnose and to distinguish from other causes (including recurrence of malignancy in some instances), now is the time for health systems to prepare for increased long-term needs to monitor and manage such treatment-related conditions.

Predicting and monitoring treatment - biomarkers

A major hurdle to a wider and more successful use of immunotherapies is the low response rate to treatment; only a small sub-group of patients benefits from immunotherapies. For example, response rates as low as 20 % have been reported for oral squamous cell carcinoma. Currently there are no biomarkers known for the prediction of responsive conditions, or the monitoring of treatment over time. This is obviously a subject of intense research.

One biomarker that has been explored in some detail is the PD-L1 ligand (which activates the T-cell inhibiting PD-1 receptor) but it does not provide reliable monitoring or prediction. It has been noted that a high concentration of tumour-infiltrating cells (a so-called immune-inflamed tumour profile) correlates with better survival, and so specific cells might be monitored as indicators of response to treatment. However, it is not clear if this characteristic also correlates with response to immunotherapies. A recent suggestion is based on the observation that especially tumour types with a greater mutation burden (melanoma, high mutation rates when associated with UV-exposure; or some types of lung cancer, high mutation rates when associated with smoking) tend to respond well to immunotherapies, suggesting that the mutation burden of a tumour and/or antigens could serve as biomarkers.

For the time being, no biomarkers that could help to identify and predict response to immunotherapies have been identified. Numerous ideas and hypotheses are being explored. The currently available information strongly suggests that there will not be a single predictive biomarker, but that any reliable predictions will have to simultaneously consider different components. This is likely to include DNA sequencing of the tumour to identify specific genetic mutations, RNA sequencing for obtaining finger-printing of the immune-system phenotype, PD-L1 and/or other ligand and/or antigen characterisation as descriptors of the tumour microenvironment. Furthermore, such investigations aim at a moving target as both host physiology and tumour metabolism co-evolve over time, thus repeated investigations will be needed. If and how such requirements will be practical and can be embedded in clinical practice remains to be seen.  

Some specific immunotherapy aspects of head & neck malignancies

In trying to relate this promise of an effective new treatment to head & neck cancers there are some areas of promise. The search for biomarkers actually demonstrates that strong immunomodulatory / immunosuppressive characteristics are typical. These may make good candidates for immunotherapy when targeting / activating the PD-1 receptor or its ligand PD-L1 (where present / possible).

Pronounced escapism from immune surveillance may be related to anatomical location in the midst of a dense lymphatic network (the logic being that only tumours that are already good at hiding from immune surveillance have a chance to develop / grow in this location) / may make for a particular tumour microenvironment / may account for a particularly dynamic evolution of tumour metabolism.

One can see the immediate problem in relation to the number of times the word ‘may’ has been used in the above paragraph.

Head and neck cancers are extremely heterogeneous conditions (and certainly not simply HPV related in the way most cervical cancers are), they vary in behaviour not only in histopathological terms but in relation to specific anatomical sites. Therefore they break down to small numbers of closely related tumour types. This means that it is not likely that large clinical trials will be feasible for reasonable narrowly defined conditions (see discourse above), the more feasible approach will be careful observational studies.

Some rather specific immunosuppressive properties have been reported where there seem to be some correlations between immunosuppressive characteristics and particular genes / mutations. It is not yet clear what (if any) the causal relationships are. Tumour genomics is, therefore particularly high up on the to-do list of research into this kind / location of carcinoma.

In addition, very practical ethical considerations will apply. There have been suggestions that pre-treatment with low dose radio/chemotherapy promoting tumour bed inflammation and the mutations induced by these may make ‘dynamically evolving’ tumours more susceptible to immunotherapies. This would be the ethical equivalent of delivering sub-optimal treatment to malignancies potentially curable by conventional means in order to conduct a prospective randomised clinical trial of something that may – or may not be ‘better’ (in whatever way the trial designers decide).

At the time of writing, immunotherapy for the curative treatment of head and neck cancers is a source of fascination for scientists and possibly concern for clinicians. It seems unlikely to be of relevance to most head and neck cancer patients in the immediate future without that dichotomy being fully addressed and resolved.