The Integration of Radiotherapy with Immunotherapy for the Treatment of Non–Small Cell Lung Cancer

Lung cancer is the second most common cancer in the United States (1), and non–small cell lung cancer (NSCLC) comprises the majority of cases (2). Five-year survival rates range from 14% to 49% for node-negative NSCLC, and <5% for locally advanced, node-positive through metastatic disease (3). Surgical resection is often the preferred treatment approach for early-stage NSCLC (4), and radiotherapy (RT) is a mainstay of treatment for early-stage NSCLC patients considered inoperable. Stereotactic body radiation therapy (SBRT), also known as stereotactic ablative RT (SABR), can be safely and effectively delivered to patients with inoperable early-stage disease, including those with poor pulmonary function at baseline (5). RT may be given concurrently or sequentially with chemotherapy in patients with locally advanced disease (4). Systemic therapy with platinum-based combination chemotherapy, or an agent directed to either targetable rearrangement/mutation (such as ALK, ROS1, or EGFR) or other targets [such as programmed cell death-1 (PD-1)], is recommended as first-line treatment in patients with metastatic disease (4).

Despite therapeutic advances, there remains a significant need to improve outcomes across all stages. Although SBRT provides excellent local control rates (98% at 3 years and 87% at 5 years) in patients with early-stage NSCLC (6, 7), patients often develop distant metastases in follow-up, suggesting an ongoing need to improve systemic disease control, even in earlier stages. In the setting of unresectable stage III NSCLC, trials of induction or consolidation chemotherapy with concurrent chemoradiation have failed to show improved tumor control or survival benefits beyond standard-of-care chemoradiation (8, 9). Immunotherapy with checkpoint blockers has been a recent addition to the armamentarium of NSCLC treatment for patients with metastatic disease (10–12), and for patients with unresectable stage III disease with no evidence of disease progression after concurrent platinum-based chemoradiation (13).

There is now increasing recognition of the complex interplay between RT and the immune system, with a greater appreciation of the ability of RT to influence systemic tumor responses. This has led to significant interest in approaches that combine RT with immunotherapy, harnessing both components to expand the currently available therapeutic options. In this review, we describe the current practices in utilizing RT and immunotherapy for NSCLC, present the rationale and early results of integrating these modalities, and highlight some of the key unanswered questions in the field.

For early-stage node-negative NSCLC, surgery is recommended for patients with operable disease, whereas SBRT is the recommended approach for patients considered medically inoperable (4). In retrospective studies of patients with early-stage NSCLC, SBRT has been shown to provide similar rates of local control and cancer-specific survival in inoperable patients as for patients undergoing surgical resection (14, 15). A pooled analysis of 58 medically operable stage I NSCLC patients from two phase III trials, STARS and ROSEL (which did not meet their accrual goals), suggested that SBRT is better tolerated and may lead to improved OS compared with surgery (16).

For patients with resectable locally advanced NSCLC, the standard treatment is surgery followed by adjuvant chemotherapy (4). Adjuvant RT is indicated in patients with mediastinal nodal involvement and/or positive resection margins, with RT offered either sequentially or concurrently with adjuvant chemotherapy (4). Neoadjuvant chemoradiation and chemotherapy prior to surgery are potential treatment options for select patients (4). In patients with locally advanced NSCLC who cannot undergo surgery, concurrent chemoradiation with conventionally fractionated RT is typically recommended (4, 17). A meta-analysis of 25 randomized clinical trials showed significantly decreased risk of death with concurrent chemoradiation versus RT alone [hazard ratio (HR), 0.71; 95% confidence interval (CI), 0.64−0.80] and versus sequential chemoradiation (HR, 0.74; 95% CI, 0.62−0.89) in patients with stage III NSCLC (17).

Failure pattern analyses have shown that not only are locally advanced NSCLC patients at a significant risk of developing distant metastases, but a notable proportion of recurrent patients also show locoregional failure (18). Thus, defining the RT dose and fractionation for optimal locoregional tumor control remains an important area of investigation. The role of RT dose escalation was addressed by RTOG 0617, a phase III study of patients with unresectable stage IIIA/B NSCLC that demonstrated that concurrent chemoradiation with high-dose RT (74 Gy) was associated with decreased overall survival (OS) and increased incidence of treatment-related deaths and severe esophagitis versus chemoradiation with standard dose RT (60 Gy; ref. 19). Consequently, based on current evidence, conventionally fractionated RT dose beyond 60 Gy should generally be avoided in the setting of concurrent chemoradiation. A meta-analysis of 10 randomized controlled trials observed that accelerated or hyperfractionated RT provided a modest OS benefit compared with conventional RT in locally advanced NSCLC (HR, 0.88; 95% CI, 0.80−0.97; P = 0.009; ref. 20). A study of adaptive RT, with treatment fields informed by ¹⁸F-fluorodeoxyglucose-positron emission tomography/computed tomography (FDG-PET/CT) conducted at mid-treatment, demonstrated improved locoregional tumor control in patients with unresectable or inoperable stage II–III NSCLC who were receiving concurrent chemotherapy (21). These results have provided the background for the ongoing phase II study RTOG 1106, evaluating an individualized adaptive RT plan used after FDG-PET/CT compared with standard RT in unresectable or inoperable stage III NSCLC patients who receive concurrent chemotherapy. Multiple studies have also shown a positive local control benefit of SBRT boost after concurrent chemoradiation in patients with locally advanced NSCLC, with no additional toxicity beyond what is expected for conventional RT (22–25).

Despite improvements in clinical outcomes from current treatments (SBRT in early-stage node-negative NSCLC, and concurrent chemotherapy with conventionally fractionated RT in locally advanced NSCLC), there has been only a modest improvement in endpoints including freedom from progression, recurrence, and/or metastasis as well as OS (6, 7, 14, 15, 18). These limitations provide further rationale to support studies such as the combination of RT or chemoradiation with immunotherapy.

In addition to providing effective local treatment at the irradiated tumor site, RT can also mediate an abscopal effect, a phenomenon in which tumor regression occurs in nonirradiated lesions (26–28). Although the abscopal effect has been observed for decades, the exact mechanisms underlying this phenomenon remained elusive for much of this time (27). Demaria and colleagues were the first to link the abscopal effect of RT to an immune-mediated mechanism (29), and recent studies have provided further evidence that RT has the potential to activate a tumor-directed systemic immune response (26–28). With the advent of immunotherapy, new areas of research have emerged, with significant recent interest in combining RT with immunotherapeutic agents to potentially enhance the abscopal effect (26).

Preclinical evidence points to RT as a priming event for immunotherapy. By modulating the host's immune system, RT can render tumor cells more susceptible to T-cell–mediated attack (30). RT promotes the release of tumor neoantigens from dying tumor cells, enhances MHC class I expression, and upregulates chemokines, cell-adhesion molecules, and other immunomodulatory cell surface molecules, thereby potentiating an antitumor immune response by triggering immunogenic cell death (Fig. 1; refs. 30, 31). The expression of programmed cell death ligand-1 (PD-L1), a checkpoint activated by tumor cells to evade the immune system, is upregulated in response to RT in preclinical models of NSCLC and other tumors (Fig. 1; refs. 31–34). RT and PD-L1 blockade in mouse models of melanoma, colorectal cancer, and breast cancer resulted in significantly delayed tumor growth, which was mediated by CD8⁺ T cells (32, 35). Similar synergistic antitumor activity has also been noted in mouse models of NSCLC (33, 34). Using a mouse model of colon carcinoma, Dovedi and colleagues found that concurrent but not sequential administration of RT and anti–PD-1 therapy led to tumor regression of distal nonirradiated lesions (36). In a study using a melanoma mouse model, RT given before or concurrently with CTLA-4 blockade delayed tumor growth and improved survival, compared with mice treated with anti–CTLA-4 therapy alone (37). Importantly, when RT is combined with immune checkpoint blockade, the dose per fraction of RT has been demonstrated in preclinical models to have an impact on its systemic effects. RT-induced cytoplasmic double-stranded DNA is sensed by the cyclic GMP-AMP synthase (cGAS)-stimulator of interferon genes (STING) pathway to induce IFN-I, a key mediator of dendritic cell recruitment and maturation (38–40). Vanpouille-Box and colleagues demonstrated a threshold dose per fraction of RT administered beyond which TREX1 is induced (41); as an exonuclease, TREX1 digests cytosolic double-stranded DNA, thereby removing the substrate that triggers cGAS-STING signaling. These findings have significant clinical implications in the selection of dose and fractionation when RT is combined with immunotherapy.

The advent of immunotherapy for NSCLC has provided opportunities to enhance systemic tumor control by mechanisms distinct from those that underlie the use of chemotherapy or agents that target tumor-driver mutations (Fig. 2; refs. 42, 43).

Multiple agents are under active investigation in NSCLC (Fig. 3). Checkpoint molecules are regulators of T-cell activation that deliver negative or positive regulatory signals (44), and activation of inhibitory checkpoints may enable tumors to evade the immune system (45). Within the lymph node, CTLA-4 acts as an inhibitory checkpoint on T cells (42, 44). Ipilimumab and tremelimumab are anti–CTLA-4 antibodies that prevent the binding of CTLA-4 on T cells with CD80 and CD86 on antigen-presenting cells, thereby enhancing T-cell activation (46). Ipilimumab is approved for the treatment of patients with melanoma (47), whereas tremelimumab is still investigational. Within the tumor microenvironment, PD-1 is an inhibitory checkpoint that downregulates effector T cells (45). Nivolumab and pembrolizumab are anti–PD-1 antibodies that bind and block PD-1 receptors on activated T cells, whereas durvalumab, atezolizumab, and avelumab are anti–PD-L1 antibodies that bind and block PD-L1 on tumor cells (45). Blockade of the PD-1/PD-L1 pathway enables T cells to recognize and destroy tumor cells (45). To date, only the PD-1 blockers pembrolizumab and nivolumab and the PD-L1 blocker atezolizumab are FDA-approved for the treatment of metastatic NSCLC (10–12), with PD-L1 expression as a selection criteria for treatment with pembrolizumab (11). Durvalumab has been recently approved by the FDA for the treatment of unresectable stage III NSCLC patients whose disease has not progressed following concurrent platinum-based chemoradiation (13). Avelumab is currently being tested in clinical trials in NSCLC.

Other targets for immunotherapy include OX-40 (48, 49), toll-like receptors (TLR; refs. 50, 51), IDO1 (52, 53), and cytokines (54). OX-40 agonists, TLR agonists, IDO1 inhibitors, cytokines, and cytokine blockers are currently in clinical development, and some of these agents are being tested in combination with RT.

The integration of RT with checkpoint inhibitors may present overlapping toxicities such as pneumonitis (10–12) and, potentially, myocarditis (55). Recent data from clinical trials and secondary or retrospective analyses include patients with locally advanced or metastatic NSCLC. Durvalumab is the only checkpoint blocker with phase III data following definitive platinum-based concurrent chemoradiation in locally advanced, unresectable stage III NSCLC patients without evidence of disease progression (56). The phase III randomized PACIFIC study investigated durvalumab versus placebo in 713 patients with unresectable stage III NSCLC who did not have disease progression after ≥2 cycles of platinum-based chemoradiation (56). The incidence of adverse events (AE) of any cause was similar for durvalumab versus placebo for AEs of any grade (97% vs. 95%), grade 3/4 AEs (30% vs. 26%), and serious AEs (SAEs; 29% vs. 23%; ref. 56). Treatment-related AEs of all grades were reported in 68% of durvalumab-treated patients versus 53% of placebo-treated patients, and the incidence of grade 3/4 AEs was 12% versus 4%, respectively (56). The most frequent grade 3/4 AEs of any cause were pneumonia (4% in each arm) and pneumonitis (3% in each arm), which were expected after definitive chemoradiation (56). The incidence of treatment-related grade 3/4 pneumonitis was comparable at 1% in each arm (56). A planned interim analysis of the PACIFIC trial showed that durvalumab significantly extended both the median progression-free survival (PFS) versus placebo (16.8 vs. 5.6 months; HR, 0.52; 95% CI, 0.42–0.65; P < 0.001) as well as the median time to death or distant metastasis (23.2 vs. 14.6 months; HR, 0.52; 95% CI, 0.39–0.69; P < 0.001), and significantly increased the objective response rate (28% vs. 16%; P < 0.001; ref. 56). Based on these data, durvalumab recently received FDA approval for the treatment of unresectable stage III NSCLC patients achieving at least stable disease following definitive chemoradiation (13).

In a secondary analysis of the phase I KEYNOTE-001 study, 24 of 97 patients with metastatic NSCLC received thoracic RT prior to pembrolizumab. Pulmonary toxicity of any grade was observed in 63% of patients (15/24) with previous thoracic RT versus 40% of patients (29/73) with no previous thoracic RT (P = 0.052; ref. 57). The incidence of pembrolizumab-related pulmonary toxicity of any grade was significantly higher in patients with previous thoracic RT (13%; 3/24) versus patients with no previous thoracic RT (1%; 1/73; P = 0.046; ref. 57). Pembrolizumab was associated with significantly longer PFS and OS in patients who previously received extracranial RT versus those without prior extracranial RT (PFS, 6.3 vs. 2.0 months; P = 0.008; OS, 11.6 vs. 5.3 months; P = 0.034; ref. 57). Early results from a phase II study of consolidation therapy with pembrolizumab following concurrent chemoradiation in 93 patients with unresectable stage III NSCLC showed that 15% of patients (14/93) experienced pneumonitis of grade ≥ 2, 7% (6/93) had grade 3–5 pneumonitis, and 2 patients died due to pneumonitis (58).

Multiple retrospective analyses have provided additional insight beyond recent clinical trials. In a retrospective analysis of advanced NSCLC patients (n = 201) treated with nivolumab, 50 patients (25%) had received prior thoracic RT, among whom 34 (68%) experienced RT-related pneumonitis before nivolumab (59). The incidence of nivolumab-related interstitial lung disease (ILD) was higher in patients with a history of RT-related pneumonitis versus those without such history (27% vs. 10%; P = 0.018); similarly, nivolumab-related ILD was higher in patients with prior thoracic RT versus those without prior thoracic RT (22% vs. 9%; P = 0.03; ref. 59). Of note, a history of RT-related pneumonitis versus no history was significantly correlated with improved median PFS (3.6 vs. 2.3 months; P = 0.023; ref. 59). Prior treatment with thoracic RT versus no RT did not significantly extend median PFS (3.3 vs. 2.2 months, respectively; P = 0.635; ref. 59). In a retrospective analysis of NSCLC patients (n = 146) treated with nivolumab after second-line systemic chemotherapy, no difference in PFS was found between patients who received RT 6 months before receiving nivolumab and those without previous RT (60). In a retrospective analysis of 133 patients with metastatic cancer (n = 71 with NSCLC) who received RT and immunotherapy either concurrently or sequentially (anti–PD-1 therapy, 66%; anti–CTLA-4 therapy, 21%; both anti–PD-1 and anti–CTLA-4 therapy, 13%), 35% of patients experienced immune-mediated AEs (imAEs) of any grade, which were manageable and not related to the site of RT (61). ImAEs of any grade were more frequent in patients who received both anti–CTLA-4 and anti–PD-1 therapies versus either immunotherapy alone (71% vs. 29%; P = 0.0008), and trended toward increased frequency in patients who received RT within 14 days of immunotherapy compared with >14 days (39% vs. 23%; P = 0.06; ref. 61). Taken together, these data suggest that combining RT with immunotherapy does not appear to produce significantly increased toxicities beyond those associated with each treatment independently, and that this combined treatment approach may provide improved clinical benefit.

Another important rationale for the combination of RT and immunotherapy is the potential for RT to overcome resistance to immune checkpoint blockade. Yuan and colleagues described a patient with PD-L1–negative metastatic squamous NSCLC that was refractory to nivolumab (62). After palliative RT, this patient exhibited near resolution of the irradiated primary tumor and significant regression of metastases outside the RT field, hence consistent with an abscopal effect (62). Komatsu and colleagues reported the case of a patient with metastatic NSCLC who experienced an increase in size of liver metastases after treatment with nivolumab (63). The patient was subsequently treated with RT, which led to a reduction in size of the irradiated liver metastases as well as a reduction in size of pulmonary metastases, which were outside the RT field, hence also consistent with an abscopal effect (63). These case reports point to the potential role of RT in overcoming immune resistance mechanisms in NSCLC patients when administered after an initial failure of immunotherapy.

Data for combinations of RT with other immunotherapeutic agents are also available, suggesting that there is potential clinical benefit with no significant additive toxicity when combining RT with immunotherapy. The phase III study START investigated vaccination with tecemotide as maintenance therapy after chemoradiation in unresectable stage III NSCLC patients (n = 1,513; refs. 64, 65). Treatment-related AEs of any grade were reported in 34% of tecemotide-treated patients versus 27% of placebo-treated patients, and imAEs were infrequent and not significantly different between groups (64). Treatment-related SAEs were reported in 2% of tecemotide-treated patients versus 1% of placebo-treated patients, and grade 3/4 treatment-related AEs occurred in 1% of each arm (64). Among the 1,239 patients evaluable for efficacy in the START trial, 65% of patients had prior concurrent chemoradiation and 35% had prior sequential chemoradiation (64). Although the primary endpoint (OS across all randomized patients) was not met, maintenance therapy with tecemotide was associated with significantly longer median OS versus placebo (30.8 vs. 20.6 months; HR, 0.78; 95% CI, 0.64–0.95; P = 0.016) among the 806 patients who had received prior concurrent chemoradiation (64). The phase III STOP study investigated vaccination with belagenpumatucel-L as maintenance therapy after platinum-based frontline chemotherapy with or without RT in stage III/IV NSCLC patients (n = 532; ref. 66). In this trial, 5 treatment-related SAEs were reported, 3 in the belagenpumatucel-L arm and 2 in the placebo arm (66). Although this trial did not meet its primary endpoint of OS in all randomized patients, maintenance therapy with belagenpumatucel-L was associated with significantly longer median OS than placebo (28.4 vs. 16.0 months; HR, 0.61; 95% CI, 0.38–0.96; P = 0.032) among the 161 patients who had received prior chemoradiation (66).

In a phase III study of adjuvant chemotherapy or RT followed by lymphokine-activated killer (LAK) cell adoptive immunotherapy in NSCLC patients (n = 170; 63% with stage III; 69 curative cases and 101 noncurative cases), there were no serious side effects other than fever and rigors associated with LAK cell administration (67). In this study, RT followed by LAK immunotherapy resulted in improved 9-year survival versus no adjuvant therapy, or adjuvant chemotherapy or RT alone (52% vs. 24%; P < 0.001; ref. 67).

In a phase II study investigating telomerase peptide vaccination with GV1001 after chemoradiation in inoperable stage III NSCLC patients (n = 23), no treatment-related SAEs were observed (68). Vaccination with GV1001 after chemoradiation led to specific immune responses in 16 of 20 (80%) evaluable patients, with a trend toward longer median PFS in responders when compared with nonresponders (12.2 vs. 6.0 months; P = 0.20; ref. 68). In a phase II study testing granulocyte-macrophage colony-stimulating factor (GM-CSF) with concurrent chemoradiation in 41 patients with metastatic solid tumors (n = 18 with NSCLC), grade 3/4 AEs of fatigue and hematologic findings were the most common AEs observed in 6 and 10 patients, respectively (69). One patient experienced a grade 4 SAE of pulmonary embolism that emerged during GM-CSF administration; however, no patient discontinued treatment due to toxicity (69). GM-CSF combined with concurrent chemoradiation resulted in abscopal responses in 4 of 18 NSCLC patients (69). Finally, in a phase Ib dose-escalation study investigating RT followed by the immunocytokine NHS-IL2 in metastatic NSCLC patients (n = 13) attaining disease control (at least stable disease) after first-line platinum-based chemotherapy, there were only 3 grade 3 AEs related to NHS-IL2 (70). Although no objective responses were observed in this study, long-term survival occurred in 2 of 13 patients with good performance status 4 years after the start of first-line chemotherapy, including 1 patient with long-term tumor control (70).

Planned or ongoing clinical trials investigating RT combined with immunotherapy (including concurrent and sequential approaches) in patients with NSCLC are presented in Table 1. The majority of these trials are phase I or phase II studies investigating RT combined with checkpoint blockers. The only phase III study (RTOG 3505), which is currently recruiting patients, is investigating the use of nivolumab versus placebo after concurrent chemoradiation in patients with stage III unresectable NSCLC.

Although early evidence points to the clinical viability of combining RT and immunotherapy in NSCLC, many unanswered questions remain. For example, at what stage of NSCLC is this treatment approach most effective? In addition to the metastatic and locally advanced NSCLC settings, it is conceivable that patients with early-stage disease may also potentially benefit from these treatment combinations. Patients with early-stage NSCLC still have high rates of relapse after definitive resection or SBRT, as evidenced by 5-year survival rates of 30% to 49%; thus, there continues to be an unmet need for postresection adjuvant therapy and post-SBRT maintenance/consolidation therapy (3, 71).

The optimal sequence and timing of RT and immunotherapy remain investigational. Available data in NSCLC showed clinical benefit when immunotherapy was given after RT (± chemotherapy; refs. 56, 57, 67, 68). A subgroup analysis of the PACIFIC study found that the PFS improvement in favor of durvalumab was more pronounced among patients who had their last RT dose <14 days before randomization (HR, 0.39; 95% CI, 0.26–0.58), compared with those who had their last RT dose ≥14 days before randomization (HR, 0.63; 95% CI, 0.49–0.80; ref. 56). Although provocative, this finding needs to be interpreted with caution, because variables like tumor burden or performance status may have caused a treatment delay, enriching the proportion of patients with less favorable disease among those randomized >14 days from RT completion. To date, few studies have tested concomitant administration of immunotherapy with RT (69, 72, 73), and only a few case reports have described a treatment benefit with RT after immunotherapy (62, 63). The type of immunotherapy used in combination with RT is likely to play a role in whether a concurrent or sequential strategy is more efficacious. For example, RT may elicit more effective antitumor immunity if administered concurrently with anti–CTLA-4 therapy (as both therapies act at the early stage of the antitumor immune response), whereas in other cases, RT may be more effective if administered prior to anti–PD-1/PD-L1 therapy, as PD-1/PD-L1 signaling typically acts at a later stage of the antitumor immune response. Clinical trials that evaluate the optimal therapeutic sequence are needed, as are preclinical data to better understand the underlying mechanisms at play.

There are conflicting data on changes in PD-L1 expression after chemoradiation. A small retrospective analysis in locally advanced NSCLC (n = 15) found that neoadjuvant chemoradiation led to a decrease in PD-L1 expression (74), whereas retrospective analyses in patients with other tumor types found that neoadjuvant chemoradiation increased PD-L1 expression (75, 76). It is possible that the differential effects could be accounted for by differences in tumor types, staging, and/or treatment. Additional studies are needed to clarify this issue and its therapeutic implications.

The optimal RT dose and schedule to ensure adequate priming for immunotherapy, and the extent of tumor that should be irradiated while minimizing local toxicity, remain to be defined. Preclinical studies using breast and colon carcinoma mouse models (41, 77) and a clinical study in patients with melanoma (78) suggest that fractionated dosing is preferred to a single-dose strategy to elicit an abscopal response when paired with anti–CTLA-4 antibodies. However, among fractionated strategies, it is not clear if the dosing should be modeled after ablative (SABR/SBRT) or subablative RT strategies (41). In addition, in cases where high doses of RT are warranted, the use of novel technologies that cause less damage to healthy tissues, such as proton beam RT (4), should be taken into consideration as potential combination partners for immunotherapy.

The optimal RT field size also remains undefined. For example, if RT is utilized in combination with immunotherapy in settings where the malignancy is not confined to just one region, it is not clear if all gross disease or only a specific site (e.g., the primary tumor) should be irradiated, as large-field irradiation may have a negative impact on circulating immune cells and may also increase the risk of toxicity (79).

As more clinical research on the integration of RT with immunotherapy evolves, it will be important to establish and validate predictive biomarkers to identify which patients would benefit the most from RT and immunotherapy.

Currently available data, though limited, suggest that the pairing of RT with immunotherapy may be a safe and viable treatment approach in patients with NSCLC. Ongoing clinical trials will add to the growing evidence on this promising combination.

D. Raben reports receiving consulting fees from AstraZeneca, EMD Serono, Genentech, Merck, Nanobiotix, and Suvica, and is a consultant/advisory board member for AstraZeneca and Genentech. S.C. Formenti reports receiving commercial research grants from Bristol-Myers Squibb, Eisai, Janssen, Merck, Regeneron, and Varian, and honoraria from AstraZeneca, Bristol-Myers Squibb, Dynavax, Eisai, Elekta, GlaxoSmithKline, Janssen, Merck, Regeneron, and Varian. No potential conflicts of interest were disclosed by the other author.

Medical writing support was provided by Robert Schupp, PharmD, CMPP, and Francesca Balordi, PhD, of The Lockwood Group, in accordance with Good Publication Practice (GPP3) guidelines, and funded by AstraZeneca.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.