INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Lymphangioleiomyomatosis is a rare disease of unknown etiology that affects women only and nearly always presents before menopause. It mainly affects the lungs where smooth muscle proliferation surrounds the lymphatics, small airways, and blood vessels causing progressive dyspnea, hemoptysis, pneumothorax, and chylous pleural effusions (1). Lymph node enlargement and renal angiomyolipomas can occur (2). Death usually occurs from respiratory failure, and median survival has been reported as 10 yr or less from the onset of symptoms (1, 5).

The natural history of lymphangioleiomyomatosis varies between patients, and although it is usually monitored by lung function tests there are no data on the average decline in lung function in these patients. Such data are needed to plan prospective trials of treatment, to help physicians to advise individual patients on their prognosis, and to monitor the effect of therapy.

Current treatment recommendations are based on reports, largely anecdotal, that lymphangioleiomyomatosis appears to be made worse by estrogen and helped by progesterone. Two small retrospective studies (4, 5) and one systematic review of 12 evaluable case reports (6) suggest that some patients stabilize during progesterone therapy, particularly those with a chylous pleural effusion and ascites, but that few improve. Occasional patients appear to have benefited from other treatments, including tamoxifen and oophorectomy. Interpretation of these studies needs to be cautious since many patients had had more than one treatment, the indications for starting treatment were not standardized, and the systematic review may have been influenced by publication bias of positive results. Prospective studies of treatment are needed if management is to be placed on a firmer basis and these will require objective measures of disease progression.

We therefore carried out a survey of all patients known to have lymphangioleiomyomatosis in the United Kingdom during the previous 5 yr and analyzed the rate of decline in lung function obtained from hospital notes as a measure of disease progression. We also compared the decline in lung function in premenopausal and postmenopausal women and in those receiving progesterone therapy with patients who had had no hormone treatment.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We wrote to chest physicians on the British Thoracic Society register in 1994 asking for the names of patients with lymphangioleiomyomatosis who were alive or who had died within the previous 5 yr and then asked the physicians and patients for permission to study their hospital notes. Further cases were recruited subsequently by word of mouth from physicians, pathologists, and surgeons. The study was approved by the Nottingham City Hospital research ethics committee.

The diagnosis of lymphangioleiomyomatosis was accepted when confirmed by open lung biopsy or when there was a classic computerized tomographic (CT) scan of the thorax (i.e., multiple thin-walled cysts, evenly distributed throughout the lung fields with normal intervening lung parenchyma) and a compatible clinical history that might include recurrent pneumothoraces, chylous effusions, hemoptysis, progressive dyspnea, or renal angiomyolipomata. Patients with a diagnosis of tuberous sclerosis complex were excluded. Details of the patient's illness were obtained by one of the investigators (S.J.) visiting the hospital and reviewing the hospital notes (40 cases); when this was not possible the hospital notes were sent to us (seven cases) or information was provided by the patient's physician (three cases). Details of clinical course, drug treatment, surgery, comorbid disease, pregnancy, menopause, and serial lung function, including FEV₁ and carbon monoxide transfer factor (TL_CO), were collected from hospital notes. Patients were considered to be postmenopausal if cessation of menstruation was documented or if they had had an oophorectomy. Patients younger than 45 yr of age with no available menstrual history were assumed to be premenopausal. Disease duration was calculated as months since the first symptom that could be attributed to lymphangioleiomyomatosis. For example, a patient with a pneumothorax or dyspnea 5 yr before the diagnosis of lymphangioleiomyomatosis was made would be given a disease duration of 60 mo at diagnosis.

Analysis

We analyzed change in FEV₁ (ml/yr) and TL_CO (ml/min/mm Hg/yr) over time for all data that spanned more than 3 mo using linear regression (Excel 7; Microsoft, Redmond, WA). We excluded measurements obtained after a lung transplant, during pregnancy, when a pleural effusion was present and for 6 mo after a thoracotomy and 2 mo after a pneumothorax. Treatment with a -agonist was allowed. To examine the effect of duration of follow-up we carried out a similar analysis for change in FEV₁ for patients who had data that spanned at least 1, 2, and 3 yr, respectively.

To assess the effect of progesterone treatment we compared rate of change in lung function between patients receiving progesterone and those receiving no hormone treatment according to whether they were premenopausal or postmenopausal by unpaired t test. A similar analysis was carried out to assess the effect of the menopause comparing premenopausal and postmenopausal patients according to which treatment they were receiving. In six premenopausal patients data were available before and during progesterone therapy and these data were compared by paired t test. Parametric analyses were used as the rate of decline showed a fairly normal distribution (Figure 1). Results are presented as the difference between means with 95% confidence intervals (CI). Patients were excluded from this part of the analysis if they were receiving any hormone treatment other than progesterone (including hormone replacement therapy or a contraceptive drug) and if the dose of progesterone received was less than 10 mg a day or equivalent for at least 3 mo. The t tests were taken as significant at the 5% level.

View larger version (13K): [in this window] [in a new window]   Figure 1.   The relation between the rate of decline of FEV1 (ml/yr) and (a) the number of measurements of FEV1 available for analysis and (b) the duration over which measurements were made. The dashed line represents the group mean value.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Of 57 patients identified, four did not fit our diagnostic criteria and three were excluded as they had tuberous sclerosis complex. The 50 patients remaining included 41 diagnosed by lung biopsy, 31 of whom had also had a CT scan showing typical changes of lymphangioleiomyomatosis. The other nine were diagnosed from a characteristic CT scan and compatible history, which included a chylous effusion in four patients and a renal angiomyolipoma in one; the remainder were under the care of, or had had a second opinion from, a physician with expertise in lymphangioleiomyomatosis. Mean age at onset of symptoms was 34 yr; six had died at the time of analysis. Twenty-one patients had received progesterone (12 by intramuscular injection, eight orally, one by both routes), five had had an oophorectomy (only one for lymphangioleiomyomatosis), six had had tamoxifen (two for breast cancer) and five had had a lung transplant. Twenty-three patients were receiving an inhaled -agonist. Sixteen patients were postmenopausal, including the five who had had an oophorectomy and five who developed their first symptom after the menopause.

Change in FEV₁

All patients. Forty-three patients had lung function data that fulfilled the criteria for analysis, including 10 who were postmenopausal. Regression coefficients were based on a mean (SD) of nine (5) measurements over a period of 52 (35) months. The mean (SD) rate of change in FEV₁ for the 43 patients with data for at least 3 mo was 118 (142) ml/yr. The mean fall in FEV₁ was similar (101, 101, and 108 ml/yr) for the patients who had data for at least 1, 2, and 3 yr, respectively (n = 38, 33, and 32). Patients who had had more lung function measurements or who had had measurements over a longer period of time showed slightly less variation between measurements (Figure 1), although the standard deviations for subjects with measurements spanning 1, 2, and 3 yr was still large (107, 108, and 101 ml/yr, respectively). The rate of decline in FEV₁ was not related to initial FEV₁ (r² = 0.04).

Of the 43 patients included in the overall analysis a few had data that were not suitable for subsequent analyses because menopausal status was uncertain or because they had not had a 3-mo period without treatment or receiving progesterone. On the other hand, some patients contributed twice if, for example, they had data covering 3 mo while receiving no treatment and further data while receiving progesterone for at least 3 mo.

Effect of menopausal status. Of the patients with data suitable for analysis while receiving no treatment, five were postmenopausal and 20 premenopausal (Table 1). The mean (SEM) fall in FEV₁ was 170 (63) ml/yr in the premenopausal patients compared with 86 ml/yr in the postmenopausal patients. This difference was not significant.

A similar trend was seen among the patients receiving progesterone with a lower rate of decline in the postmenopausal patients than in the premenopausal patients, but the findings were not statistically significant (Table 1).

Effect of progesterone treatment. Among the women with data when premenopausal there was no difference in age, disease duration, -agonist use, or the period over which lung function was analyzed between those with data while receiving progesterone (n = 15) and those with data while receiving no hormone treatment (n = 20), but initial lung function was significantly lower in the progesterone group (Table 2). The patients receiving progesterone had a smaller decline in FEV₁ than did those not receiving progesterone (Table 1 and Figure 2), although this was not quite significant (difference between means 123 ml/yr [95% CI, 23 to +269]). The rate of decline in FEV₁ in the progesterone-treated premenopausal patients increased with increasing duration of analysis, and after 2 yr the figures were similar to those seen in the untreated patients (Table 3).

View larger version (13K): [in this window] [in a new window]   Figure 2.   Individual regression lines for change in FEV1 over time related to time of onset of symptoms in premenopausal patients receiving no treatment (n = 20) and during progesterone treatment (n = 15).

In six premenopausal patients data were available before and after the introduction of progesterone (Figure 3). After the introduction of progesterone there was a mean reduction in the rate of decline of FEV₁ of 388 ml/yr (95% CI, 36 to 740). The size of the difference was largely due to one outlier with a very rapid decline in lung function prior to progesterone treatment; removal of this patient from the analysis resulted in a smaller but still significant difference between means of 104 ml/yr (95% CI, 7 to 201).

View larger version (13K): [in this window] [in a new window]   Figure 3.   Rate of decline in FEV1 and TLCO in five patients who had data both before and during progesterone treatment. Data from the patient with much larger changes are not shown (see text).

Among the postmenopausal patients there was also a trend toward a slower decline in FEV₁ in those receiving progesterone compared with those receiving no treatment, but the numbers were small and the findings were not significant (Table 1).

Change in TL_CO

The mean (SD) fall in TL_CO for the 33 patients for whom data were available was 0.905 (1.54) ml/min/mm Hg/yr. The regression coefficients were based on a mean (SD) of 7 (4) measurements per patient made over a period of 45 (31) mo. The pattern of change in TL_CO was broadly similar to that seen for FEV₁ (Table 1), although in the premenopausal women the decline in TL_CO was significantly less in those receiving progesterone compared with untreated patients (difference between means 1.96 ml/min/mm Hg/yr [95% CI, 0.54 to 3.38]). In the five patients with data before and after the introduction of progesterone the mean reduction in the rate of decline in TL_CO was also significant (1.16 ml/min/mm Hg/yr [95% CI, 0.16 to 2.16]) after excluding the outlying patient as described for FEV₁.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Because lymphangioleiomyomatosis is rare, there is a paucity of objective data on the rate of decline in lung function with time and on the extent to which this varies in relation to events that affect sex hormones such as the menopause or after treatment such as progesterone. Although the effect of any treatment should ideally be studied prospectively this is difficult in such a rare disease, and knowledge of the rate of decline in lung function is needed to design such studies. Knowing the average rate of decline in lung function will also enable physicians to determine whether the disease in their patient is progressing more or less rapidly than average. We set out, therefore, to determine the average rate of decline in lung function of all patients with lymphangioleiomyomatosis in the United Kingdom and we related the rate of decline to the patients' menopausal status and whether they were receiving progesterone or no hormone treatment. We studied FEV₁ and TL_CO since they were the measurements made most often and because they reflect the airway and alveolar involvement by lymphangioleiomyomatosis.

The limitations of our observational approach need to be recognized. Although this is the largest published series to date the numbers in some groups were small after categorization according to menopausal status and hormone treatment. There was some variation in the dose and route of administration of progesterone, although all had had a mean dose of 10 mg progesterone or more daily as in previous studies (5, 6). Lung function was measured in different laboratories and without standardization of bronchodilator treatment and we cannot exclude the possibility of a small pneumothorax or pleural effusion being present at the time lung function was measured. The effect of bronchodilator use is probably small since most of the regression estimates are based on at least six measurements of lung function, the number of patients using a -agonist was similar in the treated and untreated groups, and the fall in lung function over the period of assessment was large in relation to the response to a -agonist. Nevertheless, for these various reasons, the data should be seen as a first attempt to quantify the decline in lung function objectively in patients with lymphangioleiomyomatosis rather than providing definitive estimates.

The mean rate of decline in FEV₁ in all patients regardless of menopausal status or hormone treatment was 118 ml/yr. This figure appears to be reasonably robust since it changed little when the analysis was restricted to patients with a longer follow-up. Studying patients with a longer follow-up and more lung function measurements should provide more reliable estimates, although such estimates will eventually be biased by favoring survivors with a lower decline in lung function. Such an effect would be small over 2 and 3 yr, and estimates over this period of time are likely to be the most reliable. The large variation in the rate of decline in lung function between patients, including those followed for 2 and 3 yr, accords with clinical experience and with the wide variation in reported life expectancy for this condition (1, 4, 5).

Evidence from a variety of sources suggests that progression of lymphangioleiomyomatosis may be accelerated by estrogen and slowed down by progesterone (2, 7), an effect that is probably mediated through effects on airway smooth muscle relaxation or proliferation. Progesterone and estrogen receptors were found on smooth muscle cells in lymphangioleiomyomatosis tissue from some patients unlike those from normal lung (10) and progesterone treatment protects against bronchoconstriction in severe premenstrual asthma, which occurs as progesterone levels fall (11). A similar relaxant effect on airway smooth muscle might improve FEV₁ in lymphangioleiomyomatosis, but it is less likely to change TL_CO. Progesterone may inhibit cell proliferation directly as seen in cultured arterial smooth muscle cells (12) or it may inhibit the response to estrogen by reducing estrogen receptors (13). Abnormal proliferation in response to estrogen was seen in cultured cells from lymphangioleiomyomatosis lung tissue from a patient with tuberous sclerosis complex (14).

Data suggesting that lymphangioleiomyomatosis may progress less rapidly after menopause (7) is in keeping with a role for estrogen in disease progression, but survivor bias of patients with milder disease is also likely. The 10 postmenopausal women in our study had a lower rate of decline in lung function, though this was not significant compared with that in the premenopausal women, possibly because of their small numbers. Onset of symptoms of lymphangioleiomyomatosis after the menopause is rare and has been associated with exogenous estrogens (4, 8) as was the case in four of the five patients in our study. Hormone replacement therapy may result in the late presentation of otherwise quiescent disease in some patients.

Various treatments have been tried in lymphangioleiomyomatosis, including progesterone and antiestrogen measures, with progesterone being prescribed most frequently in the United Kingdom. The response has usually been assessed from anecdotal reports, although retrospective reviews have looked at change in dyspnea (4) or dyspnea and lung function (5) after different treatments in 32 and 46 patients, respectively. These reviews suggest that few patients benefit from progesterone (4, 5), but as the response was categorized as "improvement, stable, or decline," a reduced rate of decline in lung function or symptoms would have been labeled as a treatment failure. In our study patients treated with progesterone showed a smaller decline in FEV₁ and TL_CO than did those receiving no hormone treatment, although the differences were only significant for TL_CO in premenopausal patients. Interpretation of these data has to be cautious, however, for several reasons. First, when patients with shorter follow-up periods were excluded the differences between the progesterone and no treatment groups disappeared. This may be due to the smaller numbers of patients, although the data would also be compatible with progesterone causing an early but nonsustained reduction in the decline in FEV₁. Second, initial lung function was lower in the patients receiving progesterone and, although the decline in lung function did not relate to initial lung function in our patients, we cannot exclude some interaction as seen with patients with ₁-antitrypsin deficiency (15). Finally, the figures may be biased by the selective inclusion of survivors or because progesterone is more likely to be given to patients with more aggressive disease in which case the effects of progesterone are likely to be underestimated. The fact that all the patients with data before and after progesterone showed a reduced rate of decline of FEV₁ and TL_CO after treatment supports a beneficial effect from progesterone.

Thus, our data provide information on the average rate of decline in lung function in lymphangioleiomyomatosis and some support for the use of progesterone treatment. We believe that a prospective clinical trial is now needed to determine the effect of progesterone and its duration of action. Using our figures we calculate that a placebo-controlled crossover trial would require some 80 patients to have 90% power to detect a 50-ml difference in change in FEV₁ over 1 yr. Clearly this would require international collaboration, but such a trial would seem both necessary and feasible. Using lung function data to determine the average rate of progression is an approach that may be of value in other rare lung diseases.