Systemic sclerosis (SSc) is a systemic autoimmune disease characterised by vascular lesions and variable degrees of extra cellular matrix accumulation causing fibrotic degenerative changes in skin, muscles, joints and viscera.1 2 The diagnosis of SSc is a clinical diagnosis, in which the presence of skin fibrosis and digital ulcers or pitting scars is combined with complaints of Raynaud phenomenon and the detection of specific autoantibodies, resulting in fulfilment of the American College of Rheumatology (ACR) classification criteria for SSc or the criteria for early diagnosis of SSc as suggested by Leroy and Medsger.3 4 The prognosis of SSc is determined by pulmonary, cardiac and renal involvement causing a 5-year mortality rate of at least 30%.5 6 To date, evidence has been accumulating that treatment with immunosuppressive drugs in an early stage of the disease could improve survival, emphasising the importance of early diagnosis.7 Although the most common causes of death in SSc are related to lung involvement,6 8 9 the most frequently affected organ is the oesophagus. Oesophageal involvement can be detected in 75–90% of the patients.10 The oesophageal abnormalities are characterised by decreased or absent peristalsis in the distal two thirds of the oesophagus and reduced lower oesophageal sphincter pressure,11 and may lead to reflux oesophagitis, and even to strictures. The pathophysiological changes of oesophageal dysmotility in SSc are initiated by local cholinergic neural dysfunction, followed by smooth muscle atrophy and fibrosis, occurring in the early course of SSc11^–13 and occurring independently of interstitial pulmonary disease, whereas in fibrotic lung diseases gastroesophageal reflux, caused by oesophageal dysmotility is associated with pulmonary disease.14 Symptoms of oesophageal dysmotility can consist of dysphagia, heartburn, dyspepsia and chest pain, or may be absent.15 Extensive investigations have shown that the correlation between complaints and abnormal findings at the specific oesophageal investigations is poor and as many as 30–40% of patients with SSc with abnormalities of oesophageal function may be asymptomatic.16 Various investigations are available for the evaluation of the oesophageal function, such as oesophageal manometry, pH monitoring of the distal oesophagus, scintigraphy, endoscopy and video barium oesophagram.16 Manometry is the most sensitive investigation to establish oesophageal dysmotility in patients with SSc, and is generally considered the gold standard.17^–19 Oesophageal manometry, endoscopy and pH monitoring are invasive investigations and poorly tolerated by some patients, and can therefore not be used as a screening tool. Most often a video barium oesophagram is used to establish or rule out oesophageal dysfunction,16 however, the sensitivity and specificity for radiological dysmotility has been determined at 60% and 80% respectively.20 In daily practice patients with SSc and complaints of either pyrosis or dysphagia are treated with proton pump inhibitors and invasive investigations are performed only if the symptoms persist.17 Since 2001 we have performed a standardised prospective follow-up to monitor the course of the disease in patients with SSc. As part of this follow-up, pulmonary involvement is assessed annually by pulmonary function tests and a high-resolution computer tomography (HRCT) scan of the chest.21 We have often observed oesophageal dilatation on the HRCT scans in patients with SSc. This raised the question whether an open oesophagus on the HRCT scan, performed for whatever indication, could be used to predict early SSc. The primary objective of this study is to evaluate whether an “open” oesophagus on the HRCT scan of the chest points to the presence of SSc in patients evaluated because of signs of interstitial lung disease. The secondary objectives are to determine whether oesophageal dilatation is more common in different disease subsets of patients with scleroderma, to examine the relationship between an “open” oesophagus and the extent of interstitial lung disease, to evaluate the relationship between complaints of dysphagia and dilatation and to evaluate the correlation between an “open” oesophagus on the HRCT scan and barium oesophageal radiographic studies.

PATIENTS AND METHODS

To select the cases of this study we used records of patients with SSc from our prospective cohort (n = 256). This cohort consists of all patients with SSc in the care of the department of rheumatology. A list was obtained of all HRCT scans performed at the department of radiology between 1 January 2004 and 31 December 2004 for the indication “interstitial lung disease”, ordered by internists, chest doctors and rheumatologists of the Radboud University Nijmegen Medical Centre. We aimed to consecutively include 100 patients with SSc from the cohort as cases for this study. The subtype of SSc, autoantibody profile, disease duration (from onset of first non-Raynaud symptom) and clinical manifestations were recorded. From the HRCT scan list, we aimed to randomly include 100 patients as controls. Their medical records were obtained to provide the clinical diagnosis of the controls; none were diagnosed with SSc.

For the cases and the controls, the first available HRCT scan of the chest in our hospital was retrieved. All HRCT scans of the chest were obtained without intravenous contrast with the patients in supine position in suspended deep inspiration. The standard HRCT protocol was used with 1.0 mm slice thickness at 10-mm intervals. All cases were evaluated by using lung windows with window level of −450 Hounsfield units (HU) and a width of 1500 HU with 1.0 mm slice thickness at 10-mm intervals, 120 kV and 240 mAs, and high-frequency spatial reconstruction algorithm.

These HRCT scans were examined in random order by two radiologists independently from each other, blinded for the diagnosis of the patients. The images were evaluated for the presence of oesophageal dilatation and interstitial lung disease. The coronal oesophageal diameter was measured at five levels on the HRCT scan of the chest, namely the level of the origin of the great vessels, the midarch of the aorta, the main carina, the pulmonary venous confluence and 1 cm above the right dome of the diaphragm. Additionally, the widest diameter of the oesophagus was measured above and below the aortic arch (fig 1). Cases in which there was disagreement, that is different measurements ⩾4 mm supra-aortic and ⩾9 mm infra-aortic as described before as cut-off values by Pitrez et al, were discussed between the radiologists in a consensus meeting and consensus was reached in every case.22

The presence and extent of interstitial lung disease was evaluated by the method of Wells et al.23 In short, this method consists of the evaluation of the scans at five levels, the same levels used for evaluation of the oesophageal diameter. Abnormal appearances are categorised as follows: grade 0 is no ground-glass pattern (GGP) or reticular pattern (RP) present, grade 1 is GGP more extensive than RP, grade 2 is GGP and RP equally extensive, and grade 3 is RP more extensive than GGP. The Wells score is a summation of the findings of all levels and ranges between 0 (no GGP or RP present) to 15 (in all levels RP present). The results from the radiologists’ readings were correlated with the clinical diagnosis and disease characteristics as mentioned above. To compare the findings on the HRCT scans with barium enema studies, we retrieved all available barium oesophageal radiographic studies of the cases and noted whether the findings were normal or abnormal. Complaints regarding the upper gastrointestinal tract were collected from the patients’ charts.

Statistical analysis

To evaluate interobserver reliability between the radiologists in judging the HRCT scan on whether the oesophagus was open or not, the Cohen κ was calculated on the results reached before the consensus meeting. The interobserver reliability for the Wells score was calculated using an intraclass correlation coefficient. Further calculations were performed using consensus results on judging the oesophageal diameter, and the mean of the Wells score of the two radiologists. Using a two by two table, the sensitivity, specificity, positive and negative predicting values of an “open” oesophagus, for presence of SSc were calculated. The analysis were performed using previous described cut-off points; oesophageal dilatation was diagnosed if the luminal coronal diameter of the oesophagus exceeded 10 mm, whatever the content (air, liquid or solid).22

The Pearson correlation of the Wells score and an “open” oesophagus was calculated for the cases and the controls. A comparison of barium oesophageal radiographic studies, complaints of the upper gastrointestinal tract and HRCT findings was performed using two by two tables allowing evaluation of the sensitivity, specificity, positive and negative predicting values. p Values <0.05 were considered statistically significant.

RESULTS

Cases and controls

The cases consisted of 105 patients, all fulfilling the ACR preliminary classification criteria for SSc or Leroy criteria for early diagnosis of SSc;3 4 71 (68%) had the limited cutaneous form and 34 (32%) the diffuse cutaneous form, and 68 (65%) were female. The mean age at the date of the HRCT scan of the cases was 57 (range 21–80) years and the median disease duration was 5.1 (range 0.1–41.6) years from onset of the first non-Raynaud symptom (table 1).

In all, 101 patients (96%) were positive for anti-nuclear antibodies (ANA), 22 (21%) for anti-centromere antibodies (ACA) and 36 (34%) were anti-topoisomerase 1 antibody positive. The mean modified Rodnan skin score (mRss) was 10.6 (range 0–44). The clinical manifestations of SSc were upper digestive tract complaints (n = 58), digital ulcers (n = 42), pulmonary hypertension defined as mean pulmonary arterial pressure (PAP) >25 mmHg at right heart catheterisation (n = 21), restrictive lung disease defined as a vital capacity (VC) <70% of predicted (n = 21), myositis defined as creatine kinase levels twice the upper limit of normal (n = 7), history of renal crisis (n = 5) and heart disease defined as diastolic dysfunction > grade 2 and/or pericardial effusion (n = 3). The median age of the controls at the date of the HRCT scan was 53.1 (range 20–82) years and 50% was female. The diagnosis of the control group consisted of leukaemia (n = 48), chronic obstructive pulmonary disease (COPD) (n = 17), sarcoidosis (n = 8), multiple myeloma (n = 5), pulmonary fibrosis (non-specific interstitial pneumonia (n = 4) and usual interstitial pneumonia (n = 3)). In 23 cases an HRCT scan dwas ordered to evaluate opportunistic infections in immunocompromise patients, or side effects of chemotherapy in malignant diseases, and suspicion of vasculitis.

Reliability of the radiological assessments

The measure of agreement of the radiologists for the oesophageal dilatation measured as Cohen κ was 0.78; the percentage of agreement was 89%. The radiologists were asked to re-evaluate the scans in which they disagreed and to come to a consensus. The mean (SD) Wells scores for the radiologists were 4.3 (5.1) and 4.1 (4.3) respectively. The correlation between the Wells scores of the radiologists was 0.77, and the intraclass correlation coefficient was 0.76.

Does oesophageal dilatation point to SSc?

Infra-aortic oesophageal dilatation was present in 65 of the 105 patients with SSc, resulting in a prevalence of 62%. In the control patients the prevalence was 12% (n = 13/107). The sensitivity and specificity of presence of oesophagus dilatation for the diagnosis of SSc or not was 63% and 88%, respectively (table 2).

The positive predictive value of an “open” infra-aortic oesophagus for the diagnosis of SSc was 83%; the negative predictive value was 69%. The dilatation of the infra-aortic oesophagus appeared to be higher at all levels in patients with SSc compared to patients without SSc, but these differences were generally not statistically significant.

Clinical variables and oesophageal dilatation

In patients with SSc with a disease duration <2 years the prevalence of infra-aortic oesophagus dilatation was 47% (14/30 patients), and in patients with SSc with a disease duration ⩾2 years it was 68% (51/75 patients), which was not significantly different. Additionally, no significant differences were noted in the prevalence of an “open” oesophagus in different subsets of the patients, such as limited cutaneous vs diffuse cutaneous, specific antibodies positive vs negative and male vs female. No significant correlation was found between the presence of muscle disease, digital ulcers, heart disease or VC <70% of predicted and infra-aortic oesophagus dilatation.

Wells score and oesophageal dilatation

A Wells score of 3 in at least one section of the HRCT scan presenting a reticular pattern ie, fibrosis, was present in 52 scans, 25 cases and 27 controls. No association was present between these Wells scores and oesophageal dilatation, neither in the cases (p = 0.52) nor in the controls (p = 0.83).

Agreement of HRCT scan and barium oesophageal radiography

In 61 of the 105 cases, barium oesophageal radiography was performed within 4 weeks of the HRCT scans. Of those 61, 38 showed abnormal oesophageal motility. Comparing this result with the HRCT scan findings provided a sensitivity, specificity, positive predictive value and negative predictive value of 71%, 61%, 75% and 56% respectively for an “open” oesophagus on an HRCT scan, assuming that abnormal oesophageal motility at barium oesophageal radiography reflects true dysmotility. The measure of agreement of the HRCT scan for the oesophageal dilatation and the barium oesophageal radiography measured as Cohen κ was 0.34; the percentage of agreement was 71%.

Oesophageal dilatation and complaints

Of all patients with SSc, treated with or without proton pump inhibitors, 58 had complaints of dysphagia (55%). When combining the complaints and oesophageal dilatation, the positive predictive value of complaints for an open oesophagus in patients with SSc was 63%. In all, 23% of the patients with SSc had an oesophageal dilatation but no dysphagia.

DISCUSSION

In this study we tested whether oesophageal dilatation as observed on an HRCT scan of the lungs could be helpful in discriminating between a diagnosis of SSc and other diagnoses in HRCT scans ordered for the evaluation of interstitial lung disease, and whether the results of these HRCT scans agreed with the results of barium oesophageal radiography. An HRCT scan is part of our annual screening program for pulmonary involvement in patients with SSc. If the HRCT scan were to provide us with information about oesophageal dilatation, a manometric study (the gold standard for oesophageal dysfunction, which is poorly tolerated by the majority of the patients) could potentially be avoided.24

The oesophageal changes in patients with SSc occur in the infra-aortic region by fibrosis of the smooth muscle cells. The differential diagnosis of dilatation in this region is primary achalasia or secondary achalasia, the latter caused among others by gastric carcinoma infiltrating the oesophagus, lymphoma and irradiation. Pitrez et al found an optimal cut-of value of a luminal diameter of 10 mm in the infra-aortic oesophagus.22 The occurrence of infra-aortic oesophageal dilatation as measured by an HRCT scan in our sample was 62%, which is in accordance with other studies using other diagnostic techniques.11 16 18 25 No difference was found in the prevalence of an “open” oesophagus in patients with early disease and a longer disease duration. Bhalla et al found an 80% prevalence of asymptomatic oesophageal dilatation in patients with scleroderma lung disease in a retrospective study of 25 HRCT scans and suggested that HRCT scans could potentially be useful in detecting asymptomatic oesophageal dilatation in these patients.26 Our study includes patients with and without interstitial lung disease and we found a lower prevalence. In contrast to the expected result, we found no relationship between oesophageal dilatation and pulmonary fibrosis in cases and controls. Oesophageal dilatation was uncommon (12%) in the control sample that did not have SSc but were suspected of interstitial lung disease. In our study we found a positive predictive value of an “open” oesophagus for the diagnosis of SSc of 83%. Therefore, the presence of an oesophageal dilatation on an HRCT scan does not completely predict the presence of SSc, and its absence does not rule out SSc. However, it could be worthwhile if a radiologist, on noticing an oesophageal dilatation on an HRCT scan of the chest, reports this finding and perhaps even suggests the possibility of a diagnosis of SSc, enabling the referring specialist to start further diagnostic workup for SSc.

The gold standard for the evaluation of the oesophageal function is manometry, which is an invasive procedure and often uncomfortable for the patients.16 27 In clinical practice barium oesophageal radiography is often performed although the correlation of barium oesophageal radiography with manometry is unknown in patients with SSc. The results of the assessment of oesophageal dilatation on an HRCT scan of the lungs agreed fairly with the results of barium oesophageal radiography. As shown by our study and by others, complaints of dysphagia do not accurately predict the presence or absence of oesophageal dilatation, therefore technical assessment still is necessary.11 16 27

The reliability of scoring oesophageal dilatation on the HRCT scans of the lungs was good according to the percentage of agreement and Cohen κ between the two radiologists. The two radiologists also reached considerable agreement in the Wells score. Therefore, it seems reasonable that in clinical practice one radiologist may determine whether an oesophageal dilatation is present or not.

In summary, we found that the positive predictive value of oesophageal dilatation as defined as an infra-aortic luminal diameter ⩾10 mm on an HRCT scan of the chest ordered for the indication “interstitial lung disease” for the diagnosis SSc was 83%, and the negative predictive value was 69%. Thus, oesophageal dilatation as visible on an HRCT may alert doctors to look for other signs or symptoms of SSc in these patients.