Delays in managing lung cancer
- Correspondence to:
Dr A Moody
Department of Respiratory Medicine, Leeds General Infirmary, Great George Street, Leeds, West Yorkshire LS1 3EX, UK;
Efforts to reduce delays in lung cancer management should not cease even though they may not affect the prognosis
It is universally acknowledged that the prognosis of lung cancer is very poor, with overall 5 year survival figures of about 5–10% worldwide.1 What is less well recognised is that the picture has changed very little over the last 20 years, and that this is in sharp distinction to other solid tumours where not only are survival rates better—in the order of 60–90%—but they have been increasing (improving) fairly rapidly and continue to do so over a comparable time. For example, in our region of the UK, comprehensive population based registry data for 2 year survival of the 5000 patients diagnosed with lung cancer in 1999 was 13% compared with 60%, 79%, 87% and 92% for colorectal, cervix, prostate, and breast cancer, respectively.2 There is merit therefore in considering what might influence and be responsible for this poor outcome.
FACTORS AFFECTING PROGNOSIS IN LUNG CANCER
The factors which affect the prognosis in lung cancer are principally the stage and related performance status at presentation, histology (that is, the biological activity of the tumour), co-morbidity, age, sex, and the time interval between first symptom and treatment.
Some of these factors are not modifiable. In theory, however, reducing intervals between presentation and treatment might downstage patients and allow an improvement in survival. There are excellent data to show that early stage disease has better survival;3 there is less good but nevertheless fairly convincing evidence that very early stage disease (that is, small asymptomatic lesions) have an even better prognosis.
There are, of course, other ways of reducing lung cancer mortality. In the very long term, over a period of decades, prevention is clearly key. In the longer term, over a 5–10 year time span, the identification and treatment of very early asymptomatic disease offers good prospects of cure (screening).4 At the present time, however, efforts to improve lung cancer mortality have to reside in reducing delays to treatment and ensuring better access to specialist appropriate care.5 However, since the best appropriate care for cure is surgery, and this modality has not changed appreciably—although now technically more adroit— for many years,6 and because the cure rate will remain low unless a higher proportion of patients present and can be managed while at stage I and stage II, the question of interval delays and their effect on prognosis is of great importance.
STAGES IN THE DIAGNOSIS OF LUNG CANCER
In considering the life history of a tumour and points at which medical intervention can take place, four intervals are pertinent:
between first malignant change and first symptom;
between first symptom and presentation;
between first presentation and confirmation of diagnosis;
between diagnosis and staging/treatment.
Interval between first malignant change and first symptom
This is the very long asymptomatic period between the first malignant change in the bronchial epithelium and the first symptom. It is a reasonable assumption that the change from Tx to T1 and then T2 is not only a local increase in size, but an increase in metastatic potential—that is, there is a greater chance that, as the tumour enlarges, the stage will increase to N1, N2, N3 or from M0 to M1. There are reasonably good data, particularly from Japanese studies, that this is the case, certainly for tumours between 0.5 and 3 cm.7,8 For example, in a study by Oda et al8 the proportion of the 409 resected specimens which had nodal disease (N1, N2, N3) was 0% for primary tumours <10 mm in diameter, 21% for tumours of 11–20 mm, 23% for those of 21–30 mm, and 48% for tumours of >30 mm in diameter. Unfortunately, most T1 lesions are asymptomatic. Tumours can enlarge markedly within lung tissue and remain silent clinically. Many of these have metastatic potential and do metastasise when they achieve a size of about 1 cm. Thus, when the first symptoms begin there is, in practice, a high chance that the tumour will be at an advanced stage, either locally invasive T3 or T4 with nodal involvement N1-N2-N3, or it will present with a metastatic symptom such as back pain.
Interval between first symptom and presentation
The interval between the patient’s first cancer symptom and presentation (within patient delay) is currently under intense investigation as a possible target for health education action. The reasons why patients present when they do and with the symptoms they do is a highly complex phenomenon which is influenced by various factors such as age and health expectations, background symptoms, fear, and their impressions about health care.9 However, as we discuss below, it is likely that attempts to shorten this interval will increase the survival chances of only a few patients with an eventual diagnosis of lung cancer.
Interval between first presentation and confirmation of diagnosis
The third interval is between first presentation to any doctor and a confirmed diagnosis in secondary care. A considerable amount of activity is presently taking place to encourage primary care practitioners to recognise potential cancer symptoms and to expedite referral to specialists and for the specialists to rapidly diagnose and stage these patients. In the UK at present, and in parallel with other healthcare systems such as in Scandinavia, there are national recommendations for these pathways which are predicated on the assumption that reducing these intervals, as well as reducing patient distress, will improve survival.10
Interval between diagnosis and staging/treatment
The fourth interval is between a confirmed diagnosis—that is, when the patient is managed as a case of lung cancer—and staging/treatment. Once again, healthcare systems are investing considerable resources in reducing this and making it uniform. There is convincing evidence that, for some patients who are potentially curable, delays at this point can decrease their chances of survival.11 Furthermore, the exponential growth pattern of tumours suggests that stage migration—that is, the change in a tumour staging from, for example, I to II, II to III, or III to IV is likely to be a far more rapid event when the primary tumour is large or when there is early nodal disease than when the tumour is, for example, a small T1 lesion.12
These points, based upon considerations of tumour biology, suggest that survival should be improved if within-patient and within-health system delays are short.
STUDY BY MYRDAL ET AL13
The paper by Myrdal et al in this issue of Thorax13 examines the impact of delay in diagnosis and treatment on the prognosis for lung cancer patients. It retrospectively analyses Swedish registry data on patients diagnosed with non-small cell lung cancer (NSCLC) over a 5 year period. Patients were excluded if they were first diagnosed at necropsy or if they received no cancer specific treatment (38% of patients diagnosed with NSCLC in the study period). Two types of delay were studied: (1) symptom to treatment delay, defined as the length of time from first onset of symptoms to the start of treatment, and (2) hospital delay, defined as the length of time from the first hospital visit to the start of treatment. The impact of these separate delays on survival was then assessed.
The results showed that the mean first symptom to treatment delay was 5.8 months and was shortest in those patients with advanced disease (3.9 months). Only 9% of patients with stage I–II disease were treated within 3 months of first symptoms. Mean hospital delay was 2.5 months and appeared to be longer for those patients with potentially curable lung cancer, but the difference was not statistically significant. On average, treatment was started 1.4 months earlier in patients with stage IV disease than in those with stage I–II disease. Survival was negatively influenced by a short delay time between first onset of symptoms and treatment: 3 year survival was 11% for patients treated within 3 months and 35% for delays of more than 6 months. Similarly, patients with the shortest hospital delay (<30 days) had a poorer prognosis.
The main conclusions of Myrdal et al were that delays in the investigation and treatment of lung cancer exceeded the recommended time scales advised by the Swedish Lung Cancer Study Group in the majority of their patients. They state that neither patient nor hospital delay appeared to negatively influence survival. Patients with advanced tumour stage who presented and received treatment within 30 days of the first hospital visit fared less well than those with a longer delay. They also suggest that, as NSCLC tumours have both varied cell doubling times and aggressiveness, further information is needed to allow identification of patients who have tumours that would benefit particularly from prompt treatment.
In this retrospective analysis, 42% of patients had adenocarcinoma—a figure probably three times higher than in comparable UK populations. This may influence the overall natural history of this group of patients with NSCLC. The exclusion of the 190 patients (38%) who did not receive active treatment seems wrong. There is no reason to presume that, if they had not been diagnosed earlier, they would not have been suitable for inclusion: they could have been confirmed not to have had another cancer and to have a compatible clinical picture and to have died from their cancer. Recall of first symptoms is likely to be difficult for patients (as pointed out by the authors), given this serious diagnosis, and will probably add bias to the analysis: 34% of patients could not recall their first symptom. The authors make no comment about those people who presented with an incidental finding on a chest radiograph. This group is more likely to be operable (small peripheral tumour not causing any symptoms) and it would be interesting to know if stage I/II is over-represented in these patients. If so, this would introduce another bias into the analysis.
Other groups have identified similar delays in the diagnosis and treatment of NSCLC. A review of the literature by Jensen et al14 showed that the time intervals between first symptom and contacting a doctor varied widely from a median of 7 days to 6 months. Studies examining doctor (hospital) delays are difficult to compare because of the different end points used. They vary widely from 48 days (referral to treatment)15 to 189 days (first symptom to treatment or decision not to treat)16 with a median delay of only 9 days from first visit to specialist to diagnosis for one centre.16 In a UK regional study in 1995 Billing et al17 showed that the mean total delay from presentation to operation for NSCLC was 109 days, including 1 month for pre-hospital delay and 2 months for physician delay. Overall, delays in the diagnosis of lung cancer vary widely at every step on the diagnostic/treatment pathway and the paper by Myrdal et al reflects this.
The effect of delay on prognosis has also been examined before. Bozcuk et al15 studied the prognostic consequences of delays in diagnosing and treating lung cancer and, like Myrdal et al, found that treatment (hospital delay) did not affect survival, regardless of disease stage. Billing et al17 found that the length of delay did not correlate with tumour stage for potentially resectable patients. However, setting up a quick access two stop clinic in one centre led to a substantial increase in the number of patients who had successful surgical resection.18 O’Rourke and Edwards11 found that six out of 29 patients on a UK regional waiting list for radical radiotherapy developed progressive disease so that they were later deemed unsuitable for this, which implies that even a modest delay decreased their chance of cure. However, delays in palliative treatment may not always be so crucial. For asymptomatic patients with stage III/IV lung cancer, delaying palliative radiotherapy treatment did not negatively affect quality of life, symptom control, or survival.19
The trend in these surveys showing longer hospital delay for patients with early stage lung cancer seems contradictory. However, patients with advanced disease are usually easily diagnosed by pleural fluid cytology, fine needle aspiration of a lymph node, or bronchoscopy—that is, they have an easily accessible tumour. In those patients who may be operable, diagnosis, staging and work up often involve extra steps in the diagnostic pathway, thereby adding extra hospital delay time. In one study, staging for potentially operable patients required a mean of 5.1 diagnostic tests per patient (mainly to exclude metastasis) necessitating on average an extra 20 days in the diagnostic work up.20 Patients suitable for surgery also need more detailed work up which may include cardiopulmonary exercise testing, mediastinoscopy, etc.
The study by Myrdal et al suggests that increased delay (patient or hospital) has no negative influence on survival and this is probably true for the majority of patients because of the high proportion of patients who present with stage III/IV disease. However, for those with a large but potentially radically treatable tumour, delay may be crucial. It is these patients perhaps who need to be identified and “fast tracked” through the diagnostic pathway. Identifying these people is inherently difficult but should be based on tumour size and location, performance status, and lack of constitutional symptoms such as weight loss.
So, in the light of these results, should we modify present efforts to reduce delays in patients with lung cancer? As discussed earlier, the delay between appearance of the first symptom and treatment is dependent on many factors. Presentation of the patient to the GP, decision by the GP to refer or for radiography, and waiting time to see a specialist make up the steps in the delay to a first hospital visit. The symptoms of lung cancer may be vague and non-specific. A history of haemoptysis without obvious infection will generally alert both patient and clinician. Cough is a less robust symptom but significant new persistent cough (that is, more than 3 weeks) should arouse suspicion. Many patients with lung cancer have COPD, so increasing breathlessness is part of their overall clinical progression and may not be a reliable specific symptom for lung cancer. Half of all patients do not have symptoms in primary care which suggest a diagnosis of lung cancer.2 It is therefore difficult to see how earlier referral to a chest physician can be achieved. In secondary care some centres use direct referrals from a radiologist to a chest physician. At our centre patients attend the chest clinic for an exclusion radiograph at the request of the GP. A chest physician reads these every day and relevant urgent appointments and investigations are implemented directly by the physician.
To reduce hospital delay time further will require an increase in resources and the re-engineering of clinical services (for example, “one stop” clinics). However, in some patients it will still necessarily take longer to obtain the diagnosis as they may need several investigations before a positive diagnosis is made.
The paradoxical results of Myrdal’s study do not mean that we should cease our efforts to reduce delays. There are three reasons for this. Firstly, the psychological stress on patients and their families who have a possible diagnosis of lung cancer is enormous. Delays only serve to worsen this. Secondly, there is a small group of patients with potentially radically treatable disease (especially those in whom this may be a borderline decision) who may have a different outcome if delays occur. Thirdly, there should be large rewards from being able to identify patients at an early (asymptomatic) stage when radical treatment is possible.
Efforts to reduce delays in lung cancer management should not cease even though they may not affect the prognosis