Deep Inspiration Breath Hold as Primary Strategy for Locally Advanced Lung Cancer Radiotherapy

Background Lung cancer remains a major cause of cancer morbidity and mortality. In Denmark, the incidence has shown a slight, but robust decrease over the last 10 years in males, but the incidence in females continues to increase. The relative survival at 5 years is only 11% for men and 15% for women. While surgery is the treatment of choice in early stage disease, radiotherapy combined with chemotherapy is the only treatment modality with curative potential in patients with locally advanced inoperable non-small cell lung cancer (NSCLC). Unfortunately, the overall survival remains poor and loco-regional disease progression or recurrences are common. Clinical studies have indicated a radiation dose response relationship, suggesting a higher radiation dose would improve local control. However, in a recent randomized phase III trial RTOG0617, of standard-dose (60 Gy) vs. high-dose (74 Gy) concomitant chemo-radiotherapy the overall survival was lowest in the high-dose arm. While toxicity may not be the only explanation to the negative result of RTOG0617, an increased rate of high-grade toxicity in the 74 Gy arm is certainly of major concern: this trial furthermore highlights the limited current understanding of the mechanisms of treatment related toxicities and how they limit our attempts to increase local control.

Radiotherapy-related side effects are severe, yet difficult to assess and quantify. CTC AE is the general golden standard tool for measuring adverse events. However, many patients have heart and lung related comorbidity prior to the radiotherapy and it may be difficult to completely distinguish comorbidity and treatment related toxicity, even if reliable baseline data exist. The symptomatology of pneumonitis correlates poorly with radiological findings and with a decline in lung function. Radiation induced heart toxicity is disputed in this group of patients and not all physiological responses to radiotherapy are fully understood. Serious or even lethal toxicities after high dose radiation therapy are underreported in prospective trials, most likely because the mechanisms and the time span of toxicity was different from the expected. A recent retrospective study13 described a positive correlation between the delivered treatment dose and an increase in lung tissue density after treatment, measured on consecutive follow-up CT scans. No correlation to clinical outcome was made.

In conclusion, the understanding of radiation-induced pulmonary toxicity is still limited. Reliable measures of toxicity are highly warranted in order to perform prospective trials improving the therapeutic outcome of lung cancer patients.

In DIBH radiotherapy, patients hold their breath during imaging for radiotherapy planning and during radiotherapy delivery. DIBH results in profound anatomic changes: tumour motion is minimised, the heart is drawn downward and the lungs are inflated and push the healthy away from the target volume. DIBH is now the standard treatment for patients with mediastinal Hodgkin lymphoma and breast cancer. This technique is simple, cost-efficient and spares the heart and lungs without any detriment to other healthy organs or to the target. However, lung tumours represent an additional challenge as they are mobile: their daily position must be assessed reliably in order to avoid underdosing the tumour and overdosing the healthy lung The investigators recently performed a pilot study based on 17 lung cancer patients, and found that most patients were able to maintain several consecutive deep inspiration breath holds of each 20 seconds throughout the course of radiotherapy. No a correlation between the FEV1 value and the ability to perform DIBH was observed. The range of FEV1 for the 17 patients was 1.2 - 3.2 litres (44-67% of predicted value). The patients were planned in both DIBH and free breathing but treated in free breathing. DIBH plans resulted in a significantly lower lung dose. However, previously it was observed that for some selected patients with multiple targets, DIBH is not the best treatment option. This highlights the importance of a full investigation of the potential benefits and drawbacks of DIBH in lung cancer patients.

The primary aim of INHALE is to investigate the feasibility of delivering safe and precise DIBH lung cancer radiotherapy. The secondary aims of INHALE are to investigate the potential of DIBH lung cancer radiotherapy to reduce treatment toxicity and to explore reliable and logistically manageable toxicity endpoints.

Key research questions

1a. Is DIBH well tolerated by lung cancer patients throughout their treatment?

1b. Can the prescribed dose be delivered accurately and reproducibly to the target in DIBH?

1c. Is DIBH safe?

1. d. What is the overall survival and the local progression free survival for patients treated in INHALE?
2. a. Is DIBH dosimetrically beneficial for all patients?

2b. What is the extent of the dosimetric benefit of DIBH? 3a. Is it possible to measure radiation-induced lung damage (RILD) in lung cancer patients in a pragmatic way? 3b. Can image-based measures be used for the assessment of toxicity?

Methods

INHALE is primary a feasibility study. It will be based on prospective inclusion of 80 patients with locally advanced NSCLC for treatment in DIBH. All patients with NSCLC referred for concomitant or sequential chemo-radiotherapy can be included in INHALE. For the secondary toxicity research questions, a historic cohort of approximately 350 patients with locally advanced NSCLC treated with concomitant or sequential chemo-radiotherapy in free breathing from 2009-2014 will be used for comparison.

Treatment planning Imaging for the treatment planning will be performed according to local and national guidelines (www.dolg.dk) with the addition of a DIBH FDG PET/CT and two consecutive DIBH CT scans to measure reproducibility of the consecutive DIBHs. The DIBH FDG PET/CT will be acquired during the same session as the conventional FDG PET/CT (without additional use of PET-tracer).

Audio-visual respiratory coaching will be applied to train the patients to hold their breath at a comfortable and reproducible inspiration level. Video goggles will be used during planning and during treatment, if treatment is applied in DIBH.

Patients that cannot comply with treatment in DIBH will be treated in free breathing and registered as "non compliant".

For all patients a DIBH as well as a free breathing plan will be calculated using volumetric modulated arc therapy technique, with constraints according to the national guidelines. The plan with the lowest doses to lung and heart will be chosen for patient treatment.

Treatment Patients treated with conventional free breathing technique will be treated according to standard clinical procedures including daily set-up to tumour position by cone beam CT (CBCT). Patients treated with DIBH technique will be set-up with DIBH CBCT, a technique requiring 2-3 consecutive DIBHs. DIBH treatment will be delivered gated. In case a patient looses the ability to perform DIBH during the treatment course, the conventional free breathing plan will be applied for the remaining treatment fractions and the patient registered as non-compliant.

The daily position of the target volume will be monitored on the CBCT, assuring precise delivery of the prescribed dose to the target(s). In case of differential motion between the treatment targets (i.e. change in the spatial separation of the primary tumour relatively to the lymph node metastases), these images will enable us to decide whether to proceed with the treatment or re-scan the patient and adapt the treatment plan accordingly.

Follow-up Patients will be followed according to national guidelines with the addition of supplementary lung function tests. Furthermore, the patients will be asked to contact the department in case of pneumonitis symptoms, i.e. dyspnoea, cough and moderate fever. At each control visit, medication, symptoms and hospital admissions will be registered. All heart and lung related diagnoses will be registered for all participating patients during the follow-up period. Patients will be followed for five years or until death.

Safety In case of any toxicity of grade 4 or 5, patient history and treatment plans will be reviewed. If association with treatment parameters is suspected, treatment constraints will be prospectively adjusted.

Statistics The study is primary a feasibility study and for this mainly descriptive statistics will be employed. With regards to the analysis of frequency of dosimetrical benefit of DIBH, assuming an underlying benefit rate of 75%, the width of the observed 95% confidence interval based on binomial statistics is approximately 64%-84%. This is a sufficiently accurate estimate of the rate of benefit from DIBH to initiate a phase III trial and engage other centres in prospective testing.

For the secondary research questions several candidate toxicity endpoints are tested and while the study is not powered to see a difference between the candidate endpoints, choosing the endpoint with the apparent best discriminatory power for a phase III trial should be expected to be the most effective strategy ('pick the winner' analysis). However, the potential statistical advantage will be balanced against a medical priority. In this way a maximum of knowledge from the data will be gained enabling the basis to formulate well-founded hypotheses for future studies in this field.

Statistical analysis plan The resulting toxicity from DIBH will be compared to a historical cohort encompassing approximately 350 patients treated from 2009-2014 in free breathing at investigators' institution.

A Cox proportional hazards model predicting the risk of RILD will be generated. We expect the model to involve performance status, smoking status, tumour size and -location as predictors, but the analysis will be based on a model reduction technique using backwards elimination, possibly assisted by forced entry. The resulting Cox model will be used to generate expected freedom from RILD curves. The predicted freedom from RILD can now be compared to the observed Kaplan Meier curve for the trial patients. The relative risk of RILD (observed vs. expected) can then be extracted. Bootstrap resampling or a published method using SAS software will obtain confidence bands on the expected outcome curve.

Comparison with the historic cohort will be made both as an "intention to treat" analysis. The historic cohort will be identified in a database collected in a project already approved by the Danish Board of Health (sagsnr. 3-3013-569/1/) and the Danish Data Protection Agency (jour.nr. 30-1178) and administered by Lotte Nygård.

Perspectives It is expected that one or more of the toxicity endpoints in INHALE will be solid enough to serve as a primary endpoint for a phase III trial of optimized radiotherapy of locally advanced NSCLC, enabling the radiotherapy society to gain solid evidence for the use of emerging technologies.

Furthermore, DIHB holds the potential for synergetic combination with other new principles, e.g. dose painting with boost to radio-resistant areas of the tumour, as this technique requires that tumour motion is minimal in order to deliver the dose correctly.