High-dose interleukin2 – a 10-year single-site experience in the treatment of metastatic renal cell carcinoma: careful selection of patients gives an excellent outcome
The use of immunotherapy such as checkpoints inhibitors is increasingly promising in the treatment of mRCC in recent times. As we continue to explore the exciting application of newer agent such as Nivolumab, findings from our review shows that HD IL2 remains an important treatment option for this disease group. Despite being an approved treatment option for mRCC for many years, uptake of HD IL2 has been low due to the complexity of therapy (requiring expert specialist management), its significant toxicity and modest response rate in unselected population. The rise of VEGF (Vascular Endothelial Growth Factor)-targeted therapy with more predictable efficacy and manageable side effects further reduced the appeal of HD IL2 as first-line treatment in recent times. Unfortunately, the inevitable tumour resistance associated with targeted therapies meant that therapeutic intent remains palliative for the majority.
The longstanding difficulty with HD IL2 has been the lack of ability to pinpoint the subset of patients who can benefit significantly from this treatment. There were previous reports suggesting unfavourable IL2 outcome in patients with higher disease burden (more than one metastatic site), and short progression free interval of less than 1 year [16]. Other observations possibly associated with tumour response includes development of thrombocytopenia [17], thyroid dysfunction [18], rebound lymphocytosis [19], and low monocytes and granulocytes [20]. However, these are post-treatment findings and therefore not helpful in pre-treatment patient selection.
The plausible link between pathology and response shown by Upton and colleague as well as our own results has increased optimism for this challenging treatment option. The maintained high efficacy level in our post-pathology selection cohort demonstrated by this updated report continues to uphold this hypothesis. In our case series, patients with tumours that met all three pathology criteria fared extremely well, with 1 in 2 responding to treatment and 1 in 4–5 achieving CR. This significant cumulative predictive benefit based on higher number of favourable pathology criteria met was shown in our multivariate analysis. We postulate that a high level of alveolar features may be more important than other features as previously discussed. We also postulate that the presence of high solid pattern is a favourable feature. In our experience, the presence of papillary or sarcomatoid architecture of less than 10 % is not necessarily unfavourable and should not be excluded if there were also other good histological features. Based on our experience here, an upper limit of 30 % for papillary pattern may be considered an appropriate cut-off for future cases provided other features are favourable.
We note the result from the prospective biomarker validation study by the Cytokine Working Group (SELECT) which did not show any significant predictive value of their proposed pathology selection tool also termed Integrated Selection Model (ISM) [21]. Although the response rate in SELECT was considered better (ORR 25 %) than historical result, this was nevertheless lower than that predicted by the author (30–40 %) for the required statistical power of the trial. Interestingly, the published information suggested only a very low proportion (10 %) with favourable (50 %) alveolar architecture within the SELECT population, and based on our findings showing correlation of this feature to response, this could be one of the likely factor for the lower than expected response rate in SELECT thus could explain the lack of detectable effect. Importantly, the ISM did not consider solid features to be a favourable feature whereas our data convincingly show it to be one. In addition, CAIX staining positivity was integrated in the categorisation of ISM which, given its lack of predictive value shown in our report, may also have an impact on the overall outcome. Thus we feel the SELECT trial may not have used optimal selection and the result from this trial should not discourage further exploration and utilisation of tumour pathology as predictive biomarker in HD IL2.
The development of cancer immunotherapy has been quite remarkable in the last 5 years where impressive clinical benefits are seen across various tumour sites. With rising healthcare cost and emphasis on personalised care, there is an urgent need for a clinically relevant biomarker to better select patients for a particular type of treatment. Example biomarkers of interest currently includes cell surface molecules such as PDL1, CD70 and ICOS expression, tumour mutational load by exome sequencing, as well as detailed immune cells analysis by immune profiling or by TCR sequencing. As yet, none of these has been validated for clinical use and some findings have been highly inconsistent – this applies particularly to PDL1 expression where the low level of expression and the dynamic nature of its expression seem likely to limit its value. However, the greater understanding of cancer immunotherapy and the wider availability of molecular immune assessment (including mutational analysis of the tumour, TCR sequencing and potentially identification of immunogenic targets) could potentially facilitate refinement of our selection criteria to further improve the therapeutic advantage. At present, based on our maintained high response rates in the post-selection era, tumour morphology represents the most sensitive and reliable predictor of IL2 response to date and may well have similar predictive effect in other newer immunotherapies.
Additionally, there was evidence of response and survival benefit with higher number of IL2 doses received during cycle 1 in this cohort. Similar effect was also previously observed by another group using HD IL2 in malignant melanoma [17]. This is a relatively independent and thus useful variable compared to total number of overall doses or treatment cycles which is inherently dictated by response, i.e. patient who responded will naturally do better and also continue to receive more treatment. It is however possible that although all patients offered this treatment were considered fit, there may be other likely unaccounted variables that could have affected treatment tolerance or even survival. There also seems to be a loss of effect above 20 doses, although this could be a true effect, there is a high degree of uncertainty above this threshold due to the small number of patients receiving more than 20 doses. Even with such possible limitation, this remains a convincing finding, and striving for maximal tolerance within safe limits especially during first cycle appears to be a justifiable strategy to improve outcome.
The number of metastatic organ sites also had a significant bearing on response and survival in this patient series. This is in keeping with previous reports, and may be associated with increased heterogeneity allowing tumour escape or immune suppressive effects of heavy tumour burden [12, 16]. Limiting treatment only to those with 1–2 metastatic organ sites could avoid unnecessary toxicity in those who are less likely to respond.
The treatment tolerance and toxicity was within expectation. Importantly, all adverse effects were reversible (except thyroid dysfunction) and there has been no treatment related death in this cohort. In keeping with previous reports, development of immune related events may be associated with better outcome as shown by observation of good proportion of responders among those affected by these phenomena (thyroiditis, myocarditis and arthritis).