Treatment patterns, outcomes and clinical characteristics in advanced renal cell carcinoma: a real-world US study

Kidney cancer is one of the most common malignancies in the USA with renal cell carcinoma (RCC) accounting for more than 90% of kidney cancers [1]. RCC is a heterogeneous condition with several histological types identified, of which clear-cell carcinoma represents more than 80% of cases [1,2]. Epidemiological data from the USA show that kidney cancer will result in an estimated 73,750 new cases in 2020, with estimated deaths of nearly 14,830 [3]. Localized RCC is usually successfully managed with surgery. However, metastatic RCC (mRCC) is more problematic and has proved refractory to conventional chemotherapy. Survival rates in patients with localized versus mRCC have remained very different over the decades but tyrosine kinase inhibitors (TKIs) provided improved overall survival (OS) in patients with advanced RCC (aRCC). As such, TKIs have increasingly become the standard of care and benchmark for all future treatments [4–6]. TKIs targeting mainly the VEGF receptor and platelet-derived growth factor receptor were able to demonstrate a change in the natural history of metastatic kidney cancer, and are approved for use in first- and second-line settings [7]. Compared with TKI monotherapy, recent TKI monotherapy and immuno-oncology (IO) combination approvals in the first-line setting offer the possibility of further improved efficacy outcomes based on results of Phase III studies [8].

IO therapeutic options have demonstrated clinical benefits in the treatment of aRCC. Nivolumab has demonstrated clinical benefits versus everolimus [9], while pembrolizumab [10] and avelumab [11] showed efficacy and acceptable tolerability in single-arm clinical trials. These benefits continue when these agents are combined with either another immunotherapy agent or a TKI [12–15]. In previously untreated, locally advanced or mRCC, nivolumab plus ipilimumab was shown to be superior to sunitinib in terms of OS, progression-free survival (PFS) and objective response rate (ORR), although benefit was only observed in intermediate and high-risk patients [14]. Similar risk–benefit profiles were seen with the combination of pembrolizumab plus axitinib, where significantly longer OS and PFS were seen together with a higher ORR compared with sunitinib [12,15] in first-line aRCC. Finally, the combination of avelumab and axitinib showed significantly longer PFS together with higher ORR compared with sunitinib in first-line aRCC, although this combination has yet to show any benefit in OS [13].

A few prognostic models have been developed to aid prediction of outcomes in patients with aRCC. The Memorial Sloan Kettering Cancer Center (MSKCC) risk model was introduced in the era of conventional immunotherapy [16]. Developed in the era of TKIs, the International Metastatic RCC Database Consortium (IMDC) model is the most widely used risk assessment scheme in the evaluation of mRCC, assisting in treatment decisions and prognosis [17]. Patients, grouped according to the number of prognostic factors, favorable risk (zero risk factors), intermediate risk (1–2 risk factors) and poor risk (3–6 risk factors), showed median OS of 43, 25 and 8 months, respectively. This has been validated in recent clinical trials of IO therapy, and can also be applied to patients previously treated with targeted therapy [18].

Recent US FDA approvals for the use of IOs in combination (IO/IO; IO/TKI) for treatment of aRCC are within the first-line treatment setting. There is little consensus on the preferred sequencing of aRCC treatments; however, the expectation is that some of the combinations currently approved for first-line use may prove to be of benefit in a second-line setting [19,20].

The NCCN Clinical Practice Guidelines in Oncology (NCCN guidelines) [21] preferred regimens for patients with a clear-cell histology and favorable IMDC risk comprise IO/TKI combinations and TKI monotherapies. Preferred regimens are pembrolizumab plus axitinib or pazopanib monotherapy or sunitinib monotherapy. Other recommended regimens for this group of patients includes nivolumab plus ipilimumab, avelumab plus axitinib or cabozantinib monotherapy. For aRCC patients with an intermediate or poor IMDC score, NCCN guidelines’ preferred regimens comprise IO/IO combination therapy, IO/TKI combination therapy and TKI monotherapy. Preferred regimens are nivolumab plus ipilimumab, pembrolizumab plus axitinib or cabozantinib monotherapy [21]. Similar guidelines have also been published recently by the Society for Immunotherapy of Cancer [19] for the treatment of patients with aRCC and clear-cell histology.

The treatment landscape for aRCC will continue to evolve and it is important to evaluate outcomes, patient characteristics, prognostic factors and treatment patterns associated with the management of aRCC in contemporary healthcare settings. As treatment regimen choice may depend on a patient’s IMDC risk score, there is a need to evaluate how these criteria impact the current management of patients with aRCC, especially as there is comparatively little real-world data on the use and effectiveness of these relatively new treatments.

Given the rapidly evolving therapeutic landscape, including new drug approvals, the proliferation of a wide range of combination treatments and the publication of new treatment guidelines, the objective of this exploratory point-in-time study was to provide an in-depth understanding of treatment patterns, clinical outcomes (response to first-line treatment) and patient characteristics in aRCC patients treated and managed across the US healthcare setting.

Materials & methods

Study design & study population

Real-world data were collected using the Adelphi RCC Disease Specific Programme (DSP)™, which is a cross-sectional survey conducted with oncologists, nephrologists and urologists who were actively treating patients with aRCC in clinical practice. In our study, aRCC included locally advanced and metastatic forms of RCC. This survey was conducted in the USA between February and September 2019. The DSP collects data from physicians who complete a patient record form (PRF) for individual patients. The DSP approach has been used extensively across multiple therapy areas and involves a well-established methodology [22].

Physicians were identified by a local fieldwork team and those eligible had to specialize in medical oncology, nephrology or urology, be seeing ≥10 RCC (any stage) patients per month and be personally responsible for prescribing decisions. Physicians completed electronic PRFs for the next eight patients consulted who met the eligibility criteria (≥18 years of age, had a diagnosis of aRCC, were receiving active drug treatment during the study period and were not enrolled in a clinical trial). The patient population included those who were receiving aRCC first-line treatment and patients who had completed first-line treatment and were now receiving second-line or later treatment for their aRCC. There was also an option for physicians to complete additional PRFs for aRCC patients receiving, or who had received, either first-line nivolumab and ipilimumab therapy or cabozantinib treatment.

Data collection

Data collected by physicians via individual patient records included patient demographic characteristics such as age, gender and time since initial and advanced diagnosis, together with key clinical characteristics such as current symptoms and comorbidities; IMDC score; tumor, node and metastases staging; Eastern Cooperative Oncology Group (ECOG) score and Karnofsky PS score. Also collected were treatment patterns, such as TKI monotherapy, IO monotherapy, IO/IO combination therapy or other combination therapies, and associated outcomes such as complete response, partial response, stable disease or progressive disease that were based on the physician’s own clinical judgment. Missing data were not imputed; therefore, the base of patients for analysis could vary from variable to variable and is reported separately for each analysis.

Within the PRF, physicians were asked to provide the IMDC score (stated) for each patient (i.e., physician’s subjective assessment of IMDC score); this also included the option for the physician to state an unknown or not assessed IMDC score. Additionally, IMDC risk criteria were collected for each patient so that the IMDC score could be independently calculated. IMDC scores were defined according to the IMDC criteria, which use the number of prognostic factors present according to the favorable risk group (0 prognostic factors), intermediate risk group (1–2 prognostic factors) and poor risk group (3–6 prognostic factors) [12].

While primarily an exploratory survey that did not formally test a specific hypothesis, the study did collect treatment outcomes observed in real-world or routine clinical practice that included the efficacy of treatment as seen in the response to current therapy and the safety of treatment seen via reporting of tolerability and side effects.

Statistical analysis

All analyses were descriptive in nature and no formal statistical tests were conducted. Numerical values reported included sample size, number of missing responses, mean with standard deviation (SD), median with inter-quartile range or minimum and maximum. For categorical variables, the sample size, number of missing responses, and number and percentage (%) in each category were determined.

Ethical considerations

Ethical exemption was sought and granted through the Western Institutional Review Board (WIRB study number 1-1152003-1).

Results

A total of 56 physicians (32 medical oncologists, nine nephrologists and 15 urologists) provided data for a total of 687 patients. The median number of PRFs completed by each physician for this survey was 10.5, 10.0 and 10.0 for medical oncologists, nephrologists and urologists, respectively. About half (52.0%) of participating physicians were drawn from academic/cancer centers with the remainder drawn from non academic hospital settings. The proportion of physicians from academic/cancer centers varied according to specialty and included 64.4% of medical oncologists, 41.7% of nephrologists and 36.0% of urologists.

Physician-reported demographic data derived from the PRFs were available for 445 patients who were receiving first-line treatment, while a separate cohort of 230 patients comprised those who had progressed from first-line treatment and were receiving second-line treatment at the time of data collection. There were also 12 patients that had completed second-line treatment (not included in these analyses). In the analysis presented in this study, these groups are described separately as they reflect different treatment journeys within the spectrum of current therapy for aRCC. Unless otherwise stated, all analyses for the 445 patients who were receiving first-line treatment were stratified using stated IMDC risk scores and this group of patients represents the main cohort analyzed.

Demographic characteristics

In patients receiving first-line treatment (n = 445), IMDC risk prognosis according to favorable, intermediate or poor risk, was determined by the treating physician. The IMDC score was available for 271 patients; 15.5% of patients had a favorable risk, 63.1% of patients an intermediate risk and 21.4% of patients a poor risk. The IMDC risk score was not available for 39.1% of all first-line patients (5.2% of patients had unknown IMDC score while 33.9% of patients were not assessed).

The stated IMDC risk score was compared with the derived IMDC risk score, and of 445 patients receiving first-line treatment, 42.7% of patients had both a stated and derived IMDC risk score. These scores were compared, and the stated and derived IMDC risk scores aligned for nearly half of all patients (89 out of 190 patients, 46.8%), while the IMDC risk score was over-estimated (derived scores were lower) by physicians for 24 out of 190 (12.6%) patients and under-estimated (derived scores were higher) in a further 13 patients (6.8%). For a further 64 patients (33.7%), the IMDC risk score was not assessed or was unknown, and alignment between the stated and derived IMDC scores was not possible.

First-line patients had a mean (SD) age of 64.2 (10.6) years, 31% of patients were female, the mean BMI (SD) was 26.7 (4.3), and 62.0% of patients were White/Caucasian, 16.0% of patients were African–American and 9.7% of patients were Hispanic (Table 1). According to IMDC risk score, the mean (SD) age was 62.9 (11.3) years in those with favorable risk, 63.9 (9.8) years in patients with intermediate risk and 64.2 (11.4) years in patients with poor risk. The mean BMI (SD) of first-line patients varies from 25.5 (5.0) in those with favorable risk to 26.8 (4.1) in those with intermediate risk and 27.2 (5.5) in patients with poor risk. The percentage of female patients ranged from 28.6% in patients with favorable risk to 27.5 and 41.4% in patients with intermediate and poor risk, respectively.

In the overall population, which included patients whose IMDC score was unknown or not assessed, 54.4% of patients were current or ex-smokers, while 53.7% of patients were described as abstinent or infrequent drinkers with regards to alcohol consumption. The percentage of current or ex-smokers ranged from 38.1% in patients with favorable risk to 59.1 and 39.6% in patients with intermediate and poor risk, respectively. Likewise, abstinent or infrequent alcohol consumption ranged from 76.2% in patients with favorable risk to 47.4 and 72.4% in those with intermediate and poor risk, respectively. In the overall population, 22.0% of first-line patients were working full-time, while 40.0% of patients were retired with the remaining patients either working part-time, on long-term sick leave, homemakers, unemployed or of unknown employment status (Table 1). The percentage of patients working full-time declined with increasing IMDC risk score ranging from 28.6% in patients with favorable risk to 24.0 and 20.7% in those with intermediate and poor risk, respectively (Table 1).

When health insurance status was assessed, 42.7% of patients undergoing first-line treatment had private health insurance compared with 51.7% who had public insurance (other/don't know: 5.6%). These proportions were similar regardless of whether patients were stratified as favorable risk (private, 52.4%; public, 42.8%), intermediate risk (private, 43.9%; public, 52%) or poor risk (private, 50%; public, 50%, respectively) according to IMDC risk group (Table 1).

When patients with aRCC undergoing second-line treatment were stratified by treatment type, according to treatment received at first line, 34.8% of patients had private health insurance compared with 56.9% who had public insurance (other/don't know: 8.3%). These proportions were similar regardless of whether patients were prescribed TKI monotherapy (private, 32.7%; public, 58.8%), IO/IO combination therapy (private, 40.5%; public, 48.7%) or other regimens (private, 39.3%; public, 57.1%, respectively) (Supplementary Table 2).

Clinical characteristics

Clear cell carcinoma was the most common histological RCC subtype seen in first-line patients (71.2% of patients overall), although this percentage varied across IMDC risk groups from 81.0% of patients with favorable risk to 67.3 and 63.8% in patients with intermediate and poor risk, respectively. Mean (SD) time since diagnosis of aRCC was of 7.7 (10.0) months in the overall population with a trend toward shorter times since diagnosis of aRCC with increasing IMDC risk score.

In the overall population (n = 445), at the time of data collection physicians reported that the mean (SD) number of symptoms currently experienced by patients was 3.1 (3.3); common symptoms observed in more than 15% of patients included loss of appetite (28.1%), fatigue (24.5%), weight loss (23.6%), lack of energy (21.1%), nausea (19.6%), bone pain (17.3%), back pain (15.3%) and hypertension (14.6%) (see Supplementary Table 1 for more details). There was a general increase in the frequency of these symptoms with increasing IMDC risk score with the mean value (SD) ranging from 1.7 (2.3) in patients with favorable risk to 3.2 (3.4) and 4.0 (3.9) in patients with intermediate and poor risk, respectively (Table 2). There was a diverse range of concurrent comorbidities observed in patients at the time of data collection and those present in more than 10% of patients included hypertension (29.7%), hyperlipidemia (17.1%), anemia (13.7%), diabetes without chronic complications (10.6%) and depression (11.2%). The mean (SD) number of comorbidities ranged from 3.0 (1.4) in patients with a favorable risk to 4.1 (2.2) and 4.4 (3.2) in patients with intermediate and poor risk, respectively (Table 2). The pattern of comorbidities appeared similar across IMDC risk groups.

The mean (SD) number of metastatic sites grew with an increase in IMDC risk score ranging from 1.4 (0.6) in patients with favorable risk to 1.5 (0.9) and 1.9 (1.4) in patients with intermediate and poor risk, respectively. PS (ECOG score) in the overall population (n = 445) showed that 27.6% of patients had a score of 0, while 40.2% of patients had a score of 1, and 20.2% of patients had a score of 2 or more. PS was not assessed in 11.9% of patients (Table 2).

Treatment patterns

Information on 445 patients receiving first-line treatment at time of data collection showed that the most common treatment was TKI monotherapy, which was received by 51.0% of patients; the four most common agents were sunitinib (36.1%), cabozantinib (33.5%), pazopanib (16.3%) and axitinib (8.4%). The next most common treatment received by 25.8% of patients in the first-line setting was IO/IO combination therapy (100% consisted of nivolumab and ipilimumab treatment). The remaining patients received other first-line treatments (23.1%). Other first-line treatments given to patients were numerous but included IO/TKI combination therapy (6.5%), TKI combination therapy (3.8%), IO monotherapy (6.3%) and other IO combination therapy (0.7%) such as nivolumab and bevacizumab, or nivolumab, bevacizumab and pazopanib (Figure 1). Dose reductions or increases were not common in the current population of patients and for more than 90% of patients, physicians documented that there was no dose change. In patients who were still receiving first-line therapies, the ongoing mean (SD) duration of first-line treatment was 4.9 (5.7) months. Mean duration was longer for patients on TKI monotherapy at 5.8 (6.3) months compared with IO/IO combination therapy, where the mean (SD) duration was 3.5 (3.7) months.

Looking at the assessment of treatment patterns according to IMDC score, the trend showed that patients with favorable risk seemed to have the highest use of TKI monotherapy with lower levels of TKI monotherapy use seen with increasing IMDC score. In favorable risk patients (n = 42), TKI monotherapy was the most common first-line treatment (61.9%), with sunitinib (26.2%), cabozantinib and pazopanib (each 14.3%) the most used TKIs (Figure 2). IO/IO combination therapy was the next most common treatment used in 16.7% of favorable risk patients. In intermediate risk patients (n = 171), TKI monotherapy was still the most used first-line treatment in 50.9% of patients followed by IO/IO combination therapy that was used in 26.9% of patients. Poor risk patients (n = 58) generally had the highest use of an IO-based therapy either as IO/IO combination therapy (34.5% of patients) or IO monotherapy (15.5% of patients). TKI monotherapy was used in 27.6% of patients with poor IMDC risk score. In patients whose IMDC risk was unknown or not assessed (n = 174), 56.3% of patients used TKI monotherapy, while 24.1% of patients were receiving IO/IO combination therapy.

The ongoing mean (SD) duration of first-line treatment in patients with a favorable IMDC risk score was 8.4 (8.8) months and decreasing to 4.7 (5.1) and 2.7 (2.9) months in patients with intermediate and poor risk, respectively.

Outcomes

A complete response to treatment was only reported in three first-line regimens: in 9.6% of patients receiving IO/IO combination therapy, 8.8% of patients on TKI monotherapy and 6.9% of patients on IO/TKI combination therapy. For IO/IO combination therapy, these responses stemmed from a mean (SD) duration of therapy of 3.5 (3.7) months and a mean (SD) number of cycles of 3.7 (3.2) for nivolumab and 2.8 (2.0) for ipilimumab. Corresponding partial response seen in these treatment groups were 28.7% of patients on IO/IO combination therapy, 30.8% of patients on TKI monotherapy and 17.2% of patients on TKI/IO combination therapy, while 35.7% of patients on IO monotherapy also showed a partial response (Figure 3).

Looking at treatment response, the trend showed the proportion of patients with a complete or partial response was higher for patients with favorable risk compared with those with intermediate or poor risk (45.3 vs 35.7% and 27.6%, respectively). There was also a trend for a higher percentage of patients where the response to treatment was either not determined or not recorded with increasing IMDC score ranging from 23.8% in patients with favorable risk to 30.4 and 34.5% in patients with intermediate and poor risk respectively. A similar trend was also seen with TKI monotherapy and IO/IO combination therapy options. The proportion of patients with stable disease declined with an increase in IMDC risk score, and conversely, the proportion of patients with progressive disease or a clinical relapse increased with increasing IMDC risk score. In patients with unknown or unassessed IMDC risk scores, the response profile was comparable with that in the patients with intermediate or poor risk (Figure 4).

Patients receiving second-line treatment

A total of 230 patients had completed their first-line treatment and had progressed to second-line treatment at the time of data collection. These patients represented a separate cohort from those described earlier who were patients receiving first-line therapy at data collection. This group of 230 (second-line) patients had a mean (SD) age of 66.4 (11.0) years. Of these, 39.1% of patients were female, and their BMI and ethnicity were broadly similar to first-line patient demographics (see Supplementary Table 2). Most patients on second-line treatment were treated with TKI monotherapy (71.7%) as first-line treatment, while 16.1% of patients received IO/IO combination therapy and other treatment options (each used by fewer than 15 patients). The demographic parameters between the TKI monotherapy and IO/IO combination therapy groups showed that the mean (SD) age of patients treated with TKI monotherapy at first-line was 67.7 (10.5) years compared with 62.5 (11.6) years in the IO/IO combination group. This was also reflected in a trend toward younger age at diagnosis in patients treated with IO/IO combination therapy at 58.7 (10.8) years compared with 65.7 (9.8) years in patients treated with TKI monotherapy at first line.

As patients on second-line treatment have lived with advanced disease for longer, this was reflected in the mean (SD) time since initial diagnosis in second-line patients, which was 28.7 (36.8) months, 32.5 (41.5) months for patients who received TKI monotherapy at first line and 20.4 (15.8) months for patients who received IO/IO combination therapy at first line. The most recent ECOG score in the overall population of second-line patients showed that 18.7% of patients had a score of 0, which was a small decrease in the percentage of patients with a score of 0 from that seen in first-line treatment patients, while 39.6% of patients had a score of 1 (Supplementary Table 3).

The most common regimens received at second-line were TKI monotherapy (37.4%) or IO monotherapy (33.5%). IO/IO combination therapy (12.2%) was the only other regimen to be received by more than 10% of patients. Other second-line treatments reported by physicians included TKI combinations, IO/TKI combination therapy and IO combination therapy. In patients who had received TKI monotherapy at first-line treatment (n = 165), the therapies used in the second-line setting included IO monotherapy (46.7%) or a different TKI monotherapy (26.1%). In patients who received IO/IO combination therapy at first-line (n = 37), 86.5% of patients received TKI monotherapy as their second-line treatment. The most common reason for discontinuation of first-line treatment was disease progression (52.6%), which was experienced by more patients that received TKI monotherapy at first-line (55.8%) than IO/IO combination therapy (51.4%). Only 33.0% of patients completed their first-line regimen in full.

Patients on second-line treatment had been on first-line treatment for a mean (SD) duration of 10.3 (9.4) months. Current second-line patients who received TKI monotherapy at first-line had a mean (SD) treatment duration of 11.3 (10.0) months, while mean (SD) treatment duration for IO/IO combination therapy was 6.1 (4.0) months. This was reflected in the mean (SD) time to second-line treatment, which was 13.7 (21.1) months in the overall population and 15.3 (23.3) and 8.3 (7.4) months in the TKI monotherapy and IO/IO combination therapy groups, respectively. The mean (SD) treatment-free interval was 3.6 (18.6) months for patients who had completed first-line treatment, and 1.5 (3.5) and 13.9 (42.9) months for patients receiving TKI monotherapy and IO/IO combination therapy at second-line, respectively.

Of these patients, 4.8% of patients stopped their first-line treatment due to tolerability issues. Within these patients, side effects experienced were diverse but the most common were diarrhea (54.5%), colitis (45.5%), rash, dry skin, itching or erythema (45.5%) and fatigue (45.5%). As the base size was very small, it was not realistic to compare safety profiles by type of first-line treatment.

At the time of data collection, side effects occurred in 18.1% of patients who had completed first-line treatment, while most patients (73.1%) did not experience side effects. The side-effect profile was unknown in a further 8.8% of patients. In these second-line patients, the most commonly seen side effects (in more than 20% of patients) were nausea (56.1%), fatigue (51.2%), vomiting (26.8%), decreased appetite (24.4%) and headaches/migraines (24.4%). In this group, 26.8% had been treated with an IO/IO combination therapy at first-line compared with 56.1% who had received TKI monotherapy.

Discussion

This real-world analysis of patients receiving first-line treatment for aRCC in US healthcare settings (between February and September 2019) described the treatment pattern, patient demographics and outcomes at a time when the treatment landscape was rapidly changing with the implementation of IO/TKI combination therapy as a guideline-recommended treatment option for first-line aRCC patients regardless of IMDC risk group. We found that over 75% of first-line patients received either TKI monotherapy or IO/IO combination therapy: TKI monotherapy was received by just over half of all patients, with a further quarter of patients receiving IO/IO combination therapy at time of data collection. TKI monotherapy was the prevalent first-line treatment irrespective of IMDC risk group, and just over one-third of patients treated with TKIs at first-line received sunitinib and one-third of patients received cabozantinib. Given the relatively recent approval of cabozantinib (December 2017), the level of use showed a rapid adoption by treating physicians compared with sunitinib and pazopanib, which were approved in 2006 and 2009, respectively. For cabozantinib, the demonstration of significantly improved PFS and ORR in clinical studies of patients with intermediate or poor IMDC risk when compared with sunitinib (i.e., the previous standard of care) has been an important driver of its adoption in clinical practice [23].

Current NCCN guidelines recommend IO/IO combination therapy as a preferred regimen for patients with an intermediate or poor IMDC risk [21]. This recommendation was based on the results of CheckMate 214, a randomized open-label Phase III trial that showed significant improvements in OS when compared with sunitinib [14]. In the current analysis, at the time of data collection, patients with intermediate risk were nearly twice as likely to receive TKI monotherapy than IO/IO combination therapy (50.9 vs 26.9%), whereas in patients with poor risk, the respective values were 27.6% (TKI monotherapy) and 34.5% (IO/IO combination therapy). There was a comparative underuse of IO/IO combination therapy – a therapeutic regimen shown to be superior to TKI monotherapy in patients with intermediate/poor IMDC risk. Given that this study was undertaken at a time when the treatment landscape was rapidly changing with contemporary approvals of new aRCC treatment options, there is the potential for time sensitivity in treatment patterns. No adjustments were made to account for this possible impact and the data represent a descriptive analysis only.

Although the IMDC risk classification has been widely used in studies involving IO therapy it was not designed as a prognostic indicator of this group of patients and recent consensus statements have questioned the applicability of the IMDC risk assessment in the IO era of treatment [19]. This study also appears to show that first-line patients with favorable IMDC risk had higher use of TKI monotherapy than patients with intermediate or poor risk. Around 17% of patients with favorable risk were treated with IO/IO combination therapy. Since this combination therapy was entirely nivolumab plus ipilimumab, the use in this risk group was contrary to its labeled indication and may represent limitation associated with the use of the IMDC score in patients treated with IO/IO combination therapy. In addition, another possible explanation for this occurrence may be that current NCCN guidelines recommend IO/IO combination therapy as an alternative to the preferred regimens among favorable risk patients.

In 271 patients in whom the IMDC score was known, 15.5% of patients had a favorable risk, 63.1% of patients had an intermediate risk and 21.4% of patients had a poor risk. This distribution of risk scores is comparable with that seen in other modern real-world RCC populations in both first- and second-line settings [24–26], and perhaps of greater concern is that almost 40% of first-line aRCC patients were either not assessed for IMDC risk score or had an unknown IMDC risk score. This figure was composed primarily of patients in whom the IMDC risk score was not assessed (34%), rather than the data being unavailable or not recorded. The treatment patterns for these patients tended to resemble those given for patients with favorable risk. Unfortunately, outcomes seen in this group of patients in response to first-line treatment resembled those shown by intermediate and poor-risk patients, suggesting that this group of ‘unassessed/unknown’ patients was receiving sub-optimal care. The IMDC model was validated in the era of targeted therapy and currently remains the most widely used risk assessment model in clinical practice [17,18]. Evidence that around 40% of patients were not assigned an IMDC score is therefore problematic to the provision of optimal cancer treatment and its importance when estimating outcomes in first- or second-line settings. Recent consensus statements have questioned the applicability of the IMDC risk assessment in the IO era of treatment [19].

In keeping with most RCC populations, clear cell RCC was the predominant histologic type in the cohort of first-line patients in this study. The incidence of non clear cell carcinoma (multiple histologic type) increased with IMDC score in this study, ranging from 16.7% in patients with favorable risk to 29.8 and 34.5% in patients with intermediate and poor risk, respectively. There was an indication that poorer prognoses seen through IMDC scores were associated with non-clear-cell histology, a trend seen in retrospective analyses of multiple Phase II/III studies [27,28], and this was accompanied by shorter times since diagnosis in each of the IMDC risk groups.

Assessment of response to real-world treatment with either TKI monotherapy or IO/IO combination therapy showed comparable results to those seen in Phase III clinical studies. In this study, a complete or partial response to therapy in the overall first-line treatment population was seen in 39.6 and 38.3% of patients, respectively. In comparison, the ORR seen over extended periods with nivolumab and ipilimumab combination compared with sunitinib in patients with intermediate/poor risk were 41 and 34%, respectively [14].

Newly introduced IO/TKI combination therapy, approved in 2019 during the period of this study, provides alternative treatment options that may allow clinicians to offer patients further personalized treatment according to their demographics, clinical characteristics and risk profile. While approval of this regimen overlapped with the current study and may have restricted use of this therapy, a small percentage of first-line patients (∼5%) were treated with IO/TKI combination therapy and there is an expectation that use of this combination will grow with wider acceptance, incorporation of these agents into treatment algorithms and knowledge of the relevant Phase III study data [12,13,15]. Here, the use of IO/TKI combination therapy did not change with IMDC risk score. With new therapeutic options for patients with aRCC the treatment landscape is rapidly changing, and this study provides a snapshot of current treatment patterns in first-line patients in 2019, before the widespread adoption of IO/TKI combination therapy. Future work will require the real-world evaluation of treatment patterns and outcomes in patients treated primarily with IO/TKI combination therapy in accordance with current guidelines.

In this study, the most common reason for stopping first-line treatment was disease progression (52.6%). Most patients received TKI monotherapy or IO/IO combination therapy in the first-line setting, thus levels of progression while in receipt of first-line treatment may change with the use of newer treatment options. With the introduction of new therapies, such as IO/TKI combination therapy, there are new strategies available in terms of treatment sequence. In this study, in the second-line setting, the most common regimens were TKI monotherapy and IO monotherapy. In the first-line setting, most of these patients received TKI monotherapy or IO/IO combination therapy. It is likely that IO/TKI combinations will become a preferred option in first-line settings based on efficacy seen in randomized clinical trials. However, use of IO/TKI combination therapy in the current study was low and future real-world studies should determine the optimum treatment based around use of IO/TKI combination therapy in this group of patients.

Limitations

This study was conducted in the US between February and September 2019, and the results reflect clinical practice at that time period. Given the rapidly changing treatment landscape in aRCC, it should be recognized that this represents a limitation. However, the data reported within are still valid and relevant to clinical practice. There are several reasons why the use of new therapeutic options may lag clinical guidelines. Lack of physician awareness of new guidelines or historical prescribing preferences may play a role in the time taken for adoption. Drivers of treatment choice include OS, PFS, side effects and response rates but also physician comfort in prescribing treatments which can vary between physician specialty. Another issue could be the financial aspect of new costly treatments, in that funding for use or health insurance approval for reimbursement may not yet be in place in a particular jurisdiction, indicating health economics may also be a factor. However, when we investigated patient insurance status, we found similar trends in the proportion of patients who had private or public health insurance, when stratified by both IMDC risk group and by treatment received at first line, indicating that this was not a barrier to access in this particular subset of patients. All of these are common and valid themes in real-world clinical practice, and worth reiterating.

The DSP approach to collecting data has limitations largely based around the quasi-randomized sampling and the collection of data from consecutive patients who meet specific inclusion criteria. Such patients may not be fully representative of the overall population of aRCC patients, as patients who consult frequently are more likely to be included in the sample. Due to the way the data were collected, we were unable to look at differences between patents treated according to the new treatment guidelines or not. Further selection bias can arise for physician-reported data via the absence of formalized data verification procedures, while the diagnosis is often based on the judgement and diagnostic skills of the respondent physician and a formalized diagnostic checklist is not mandated as part of the DSP methodology. For example, response to treatment was based on physician-assessed response and did not include RECIST-type criteria. This will inevitably represent one limitation in this area. Despite this potential bias, the approach is consistent with diagnostic decisions made by physicians in routine clinical practice and is, therefore, reflective of the real-world treatment of aRCC. Some data were unavailable due to missing or incomplete responses. Missing data were not imputed; therefore, the base of patients for analysis could vary from variable to variable and is reported separately for each analysis. Recall bias may arise where physicians do not have the relevant data documented within patient medical files. There are minimal inclusion and exclusion criteria applied to patients making this study representative of real-world treatment. The inclusion criteria for physicians are mainly focused on the minimum number of patients seen per month, involvement in patient management and treatment decisions. These physicians may not be representative of all treating physicians of aRCC, while inclusion is also dependent on willingness to participate in the DSP.

Conclusion

Current guidelines for the treatment of aRCC recommend IO/IO combination therapy as a preferred regimen for patients with intermediate or poor IMDC risk. This study showed that these guidelines were not widely followed in routine clinical practice with (at the time of data collection) more intermediate IMDC risk patients receiving TKI monotherapy than IO/IO combination therapy. Additionally, almost 40% of patients with aRCC on first-line treatment were not routinely assessed for IMDC risk score and, therefore, may have received sub-optimal therapy. In around half of patients, progression to second-line treatment occurred within 1 year and many experienced poor clinical outcomes with first-line treatment. There is a need for more effective first-line treatment and perhaps increased physician adherence to guideline-recommended treatment regimens. An important research question remains as to whether progression from first- to second-line treatment differs based on regimen received in the first-line setting.

While the approval of new IO/TKI combination therapies has broadened the therapeutic landscape for first-line treatment of aRCC, the current analysis suggests that the proportion of patients receiving IO/TKI combinations in the first-line setting remained relatively low at the time of the study. Further research is needed to update treatment patterns, and to determine the optimum treatment sequence in patients and how to increase adoption of these new therapies.