Cardiovascular Mortality in Women With Obstructive Sleep Apnea With or Without Continuous Positive Airway Pressure Treatment

TO THE EDITOR: In 1980, the Coronary Drug Project (1) showed that among patients assigned to placebo, those who faithfully (at least 80% of the time) took the lactose placebo had a roughly 50% reduction in death compared with those who took the pill less faithfully. The authors note “the serious difficulty, if not impossibility, of evaluating treatment efficacy in subgroups determined by patient . . . adherence . . . .” Adherent patients seem different from nonadherent patients in ways that are difficult to characterize. Campos-Rodriguez and colleagues (2) compared cardiovascular mortality rates in women “with OSA [obstructive sleep apnea] [who] were classified as CPAP [continuous positive airway pressure]– treated (adherence 4 hours per day) or untreated (adherence 4 hours per day or not prescribed).” The authors conclude that “CPAP treatment is associated with a decrease in mortality risk.” They do not indicate how many patients in the “untreated group” had not been prescribed CPAP, and how many had been prescribed it and did not use it (2). Because analyzing patient subgroups on the basis of adherence can lead to misleading results, could the authors analyze their data in ways that do not rely on such an analysis?


IN RESPONSE:
We thank Dr. Finucane for his comments. He contends that analyzing patient subgroups on the basis of adherence can lead to misleading results. He also wonders how many patients in the untreated group had not been prescribed CPAP and how many were noncompliant.
We agree that the observational design might introduce selection bias because inadequate CPAP compliance could be a marker for generally nonadherent behavior, and it was acknowledged as one of the most important limitations of our study. Nevertheless, it cannot be taken for granted that all women who were nonadherent to CPAP were noncompliant with other treatments. A recent publication (1) concluded that adherence to cardiovascular medications was not different in patients with severe obstructive sleep apnea (OSA), regardless of whether they were treated with CPAP. On the other hand, a randomized design would have ethical implications in our study because of the presence of symptomatic patients in whom CPAP could not be withheld. In fact, 507 of 838 (60.5%) women diagnosed with OSA in our cohort had excessive daytime sleepiness, based on an Epworth score higher than 10.
In our study, CPAP was prescribed according to updated guidelines (2). These guidelines advise starting CPAP treatment if a patient has severe OSA or associated symptoms (mainly, excessive daytime sleepiness). All 95 women classified as having untreated severe OSA were offered CPAP. Eleven women declined, and the remaining 84 started treatment (and eventually showed bad adherence). Of the 167 women classified as having untreated mild to moderate OSA, only 47 (28.1%) began CPAP treatment (all of them had an Epworth score Ͼ10), whereas 120 (71.8%) were not offered CPAP.
In summary, 120 of 131 (91.6%) women who did not start CPAP treatment had mild to moderate OSA without excessive daytime sleepiness, whereas those who were prescribed CPAP but were nonadherent had severe or symptomatic OSA. Both groups were completely different regarding OSA severity and symptoms, so we believe that they cannot be analyzed separately. Had we only analyzed women who did not start CPAP therapy, only 11 cases with severe OSA would have been included, and the association between severe OSA and cardiovascular mortality could not have been assessed.

Virtual Autopsy in the Intensive Care Unit
TO THE EDITOR: The concept of a virtual autopsy as presented by Wichmann and colleagues (1) is interesting, but there are some "abnormalities" of minor clinical significance that may be detected. Figure 3 in the article is said to show a right subclavian venous catheter inadvertently placed into the neck. The image shows that the course of the right-sided catheter is totally above the clavicle, thus suggesting that it actually is a right-sided internal jugular venous catheter. That the distal tip of the catheter was placed into the cervical portion of the internal jugular vein is not surprising or particularly dangerous and should not be considered a "major finding." Such placement can be easily detected with ultrasonography at the time of the procedure, rather than with computed tomography. IN RESPONSE: We appreciate the comments from Dr. Kirsch and agree that, especially for cervical central lines, ultrasonography is an easy-to-perform, noninvasive method to control the correct positioning of the device. This case once more underscores the value of virtual and classic autopsy forms for quality control (1), because control mechanisms were not sufficient to prevent the displacement of the device. This is why we disagree with Dr. Kirsch in classifying the event as minor. In a patient with septic shock, measuring central venous pressure and providing adequate volume and vasopressor therapy are considered crucial (2). Because this cannot be assured, we classified the central venous line placed into the sigmoid sinus as a "new major finding."

Successful Treatment of Diffuse Pulmonary Lymphangiomatosis With Bevacizumab
Background: Diffuse pulmonary lymphangiomatosis is a rare disease caused by uncontrolled proliferation of lymphatic vessels. Although histologically benign, it can lead to death. It is distinct from lymphangiectasis and lymphangioleiomyomatosis but is often misdiagnosed. Treatment options are limited and often ineffective (1,2).
Objective: To report successful treatment of lymphangiomatosis with bevacizumab, an antibody that blocks vascular endothelial growth factor (VEGF).
Case Report: In 2005, a 40-year-old woman presented to a hospital with chest pain and fatigue, and the physicians referred her to our hospital when they found a mediastinal mass and lesions in her spleen. Surgical exploration of her mediastinum revealed an old, capsulated hematoma without active bleeding. Mediastinal biopsies found no evidence of cancer or thoracic endometriosis, and genetic and hematologic tests excluded connective tissue disease and coagulopathies. The patient had hemoptysis and hemothorax in 2006, and surgical exploration of her thorax did not provide an explanation for the bleeding. In 2008, she began to have recurrent hemoptysis, and computed tomography showed growing masses in her mediastinum, left lung, and spleen. A revised interpretation of a previous biopsy led to the diagnosis of diffuse pulmonary lymphangiomatosis 2 years later. At that time, the patient had continuous hemoptysis with anemia. Wide dissemination of lesions precluded surgical resection, and obstruction of a bronchial artery using a coil was unsuccessful. Because VEGF is known to mediate lymphatic proliferation (3,4), we started intravenous treatment with bevacizumab, 1 mg/kg, every 3 weeks. Hemoptysis stopped, hemoglobin levels stabilized, and the tumor decreased in size ( Figure, A and B). Immunohistochemical staining showed increased VEGF-A expression in diseased lymphatic vessels compared with healthy lymphatic vessels (Figure, C), despite normal plasma levels of VEGF-A (Ͻ0.1 ng/mL). After 7 treatments, bevacizumab treatment was stopped because hypertension developed. Ten months later, our patient continued without hemoptysis, and the size of the tumor was stable on computed tomography.
Discussion: Blockade of VEGF by bevacizumab turned out to be an effective treatment for diffuse pulmonary lymphangiomatosis in our patient, which is important because treatment options are limited. In 1990, a review of lymphatic disorders (2) stated that "just as the clear fluids of the lymphatics make these vessels invisible to the naked eye, so the medical knowledge and study of the lymphatic system has been nearly invisible." During the past 2 decades, however, researchers have made considerable progress in understanding lymphatic biology, largely by identifying specific markers for lymphatic endothelium (3) and developing a better understanding of lymph vessel proliferation (that is, lymphangiogenesis). Lymphangiogenesis is mainly driven by VEGF-C and the lymph-specific receptor VEGFR-3 (3), but also by VEGF-A (4). Although therapeutic interference with VEGF-C and VEGFR-3 is still experimental, clinicians have ample experience with VEGF-A inhibition. Therefore, we treated our patient with bevacizumab, which resulted in fast reduction of tumor size and immediate clinical improvement. Another recent case report (5) suggested involvement of VEGF in lymphangiomatosis but did not provide specific evidence. Our case further supports the mediator role of VEGF-A in lymphangiomatosis by showing increased VEGF-A expression in affected lymphatic vessels (Figure, C) and tumor regression with bevacizumab ( Figure, A  and B).
Conclusion: This case contributes to the understanding of lymphangiomatosis by showing that VEGF-A is a mediator of lymphangiomatosis. Moreover, it indicates that some cases of pulmonary lymphangiomatosis can be effectively treated with bevacizumab; however, further clinical investigation is needed to determine its role in treating other cases of this disease. A. Treatment effects of bevacizumab on blood hemoglobin levels. Bevacizumab was given every 3 wk, indicated by asterisks. B. The extent of lymphangiomatosis in the left lung as visualized on chest CT 1 day before (left) and 10 weeks after (right) initiation of bevacizumab treatment. C. Immunohistochemical staining of VEGF-A in proliferative lymphatic vessels (top) vs. unaffected lymphatic vessels from the same area (bottom) (original magnification, ϫ5). The brown VEGF-A staining is evident in lymphendothelial cells of proliferative lymphangioma vessels (top, inset) (original magnification, ϫ40) but almost absent in lymphendothelial cells of the vas afferens in a lymph node (bottom, inset) (original magnification, ϫ40). In addition, the density of lymphendothelial cells is higher in lymphangioma vessels than in the vas afferens, suggesting proliferation of lymphendothelial cells. The lymphatic character of the proliferative vessels was confirmed by staining for VEGFR-3 (data not shown). CT ϭ computed tomography; VEGF ϭ vascular endothelial growth factor.