Abstract
Non-thrombotic pulmonary embolism (NTPE) is a rare but potentially fatal complication of filler injections. It can result not only from direct intravascular injection but also from the migration of fillers into veins due to local pressure. Here, we report the autopsy findings of two deaths resulting from NTPE following vaginal filler injections. The first case involved a 38-year-old woman who lost consciousness 20-40 minutes after receiving an injection of 15 mL of hyaluronic acid (HA) filler. A large amount of filler was observed in the vagina with gross embolization of the paravaginal vessels. Microscopic examination revealed HA embolism in the lungs. Despite treatment, the patient died 10 days after the procedure. The second case involved a 35-year-old woman who experienced desaturation and cardiac arrest 4 minutes after receiving a collagen filler and a hybrid filler consisting of HA and polylactic acid. An autopsy revealed NTPE and systemic embolism of the collagen filler. She died 1 month after treatment in the intensive care unit. The vagina poses a significant risk for filler injections owing to its rich venous plexus. Doctors should be fully aware of this risk, and a complete autopsy should be performed in such cases.
Introduction
Injection of soft tissue fillers has become increasingly popular worldwide. In 2023, 3,411,534 dermal filler injections were administered in the United States alone, indicating a 4% increase from 2022. It is the second most frequently performed minimally invasive cosmetic procedure after Botox [1]. Filler injections into the female genital organs such as the vagina and vulva are also gaining popularity as both doctors and patients consider it a simple and safe procedure. Although not medically indicated, many women request genital filler injections for cosmetic purposes or to improve sexual function and satisfaction [2].
Filler injections are generally associated with fewer complications than surgical vaginoplasty. However, their safety and efficacy have not been proven, and local complications, such as lumps, infections, bruising, and swelling, are relatively common [3]. Vascular complications, such as ischemic injury or embolism, are rare and can result from accidental intravascular injection or migration of the filler into the blood vessels, leading to irreversible and potentially life-threatening damage [4,5].
Many types of fillers are currently used, including hyaluronic acid (HA), collagen, and polylactic acid (PLA) [2,6]. Among these, HA is the most commonly used owing to its non-immunogenicity and degradability, which can reduce local complications. However, vascular complications can occur after HA filler injections [5,7,8]. Silicone oil, once widely used, is rarely injected because of numerous reported adverse effects, including pulmonary embolism [9-14].
Herein, we present two cases of fatal non-thrombotic pulmonary embolism (NTPE) that developed after vaginal filler injection. In both cases, NTPE was confirmed by autopsy and was determined to have contributed to the death.
Case Report
1. Case 1
A 38-year-old female was admitted to the emergency department (ED) after suddenly losing consciousness twice on her way home after receiving a vaginal filler injection at an obstetrics and gynecology (OB-GYN) clinic. The first syncope episode occurred 20-40 minutes after the procedure, and the patient reportedly experienced palpitations and dizziness before losing consciousness. She underwent the procedure four times over 7 months, with a total of 47 mL of HA filler injected into the vaginal wall. The last injection was administered 48 days previously. On the day of the collapse, an additional 15 mL was injected over 5 minutes. Her blood pressure, heart rate, respiratory rate, and oxygen saturation, as measured by the Emergency Medical Service (EMS), were 96/76 mmHg, 104 bpm, 18/min, and 80%, respectively. Upon arrival at the ED, the patient appeared dyspneic and cyanotic. Doctors initially suspected anaphylactic shock because of the rapid onset of symptoms. However, her symptoms did not improve after administration of epinephrine, methylprednisolone, and chlorpheniramine.
Laboratory examinations revealed an elevated D-dimer level of 677 ng/mL (reference range, 0 to 500 ng/mL). Over 90 minutes, creatine kinase–myocardial band levels increased from 3.12 ng/mL (reference range, 0 to 5.8 ng/mL) to 5.83 ng/mL, and troponin T levels increased from 0.121 ng/mL (reference range, 0 to 0.1 ng/mL) to 0.285 ng/mL. Electrocardiography revealed sinus tachycardia with an S1Q3T3 pattern. Right ventricular dilatation with a suspicious D-sign was observed on transthoracic echocardiography. Clinical findings suggested pulmonary embolism; therefore, chest computed tomography was performed. However, there were no filling defects in the pulmonary arteries, even though the right ventricle and pulmonary trunk were dilated.
In the ED, her dyspnea worsened, and she started showing seizure-like movements. The patient was intubated and admitted to the intensive care unit (ICU). Despite the administration of vasopressors and inotropic drugs, blood pressure decreased to 32/24 mmHg, and extracorporeal membrane oxygenation was provided. The right ventricular failure and pulmonary hypertension eventually worsened. The patient developed multiple organ failure and died 10 days after admission.
Autopsy revealed a large blood clot in the vagina. Vaginal sectioning revealed that a large amount of filler had been injected into the posterior vaginal wall. The filler extended from the vaginal submucosa to the rectal mucosa. Gross embolization of the filler into the paravaginal vessels was observed (Fig. 1A). Microscopic examination revealed an amorphous basophilic substance and hemorrhage in the vaginal and rectal walls. Multifocal mild neutrophilic and eosinophilic infiltration and a foreign body reaction surrounding the substance were observed. Embolization of the basophilic substance was observed in the large paravaginal and small submucosal veins of the vagina (Fig. 1B-D). The basophilic substances stained positive for Alcian blue.
Fig. 1.
Macroscopic and microscopic findings of the pelvic organs in case 1. (A) Hyaluronic acid filler was injected into the posterior vaginal wall. Vascular embolization is grossly visible (arrow). (B) Amorphous basophilic substances are observed in the submucosa and muscular layer of the vagina. It is accompanied by hemorrhage and emboli in the small submucosal veins (arrowheads; H&E, ×40). (C) Amorphous basophilic substance extended up to the rectal submucosa (H&E, ×40). (D) Embolism of the basophilic substance in large paravaginal veins (H&E, ×40).
The lungs were severely congested, and the cross-sections appeared hemorrhagic (Fig. 2A). The right lung weighed 1,192 g, and the left lung weighed 984 g. Microscopic examination revealed diffuse intra-alveolar hemorrhage. Multifocal amorphous basophilic emboli were observed in the interstitial vessels, and some were accompanied by foreign body reaction. The emboli stained positive for Alcian blue, suggesting that the HA filler had embolized to the lungs (Fig. 2B-D). Diffuse alveolar damage was also observed.
Fig. 2.
Macroscopic and microscopic findings of the lungs in case 1. (A) The lungs are severely congested. The cross-sections appear hemorrhagic. (B) Diffuse alveolar with multifocal basophilic emboli in the interstitial vessels (arrows; H&E, ×40). (C) Foreign body reaction around some vessels with basophilic emboli (H&E, ×200). (D) The emboli stained positive with Alcian blue (Alcian blue stain, ×200).
Microscopic examination of other organs, including the urinary bladder, heart, and kidneys, revealed mild eosinophilic infiltration. No evidence of embolism was found in any other organs. The postmortem tryptase level measured in peripheral blood was 16.5 μg/L.
2. Case 2
A 35-year-old female experienced cardiac arrest after receiving vaginal filler injections at an OB-GYN clinic. Under sedation with ketamine, midazolam, and propofol, two types of fillers were injected over 5 minutes: a collagen filler and a hybrid filler consisting of HA and PLA. The injected dose is unknown. Four minutes after completion of the procedure, her oxygen saturation started to decrease. Oxygen supplementation was not helpful, and the patient eventually developed cardiac arrest. After 1 month of conservative treatment in the ICU, the patient died of hypoxic brain damage and pneumonia.
On autopsy, discoloration of the vaginal mucosa was observed at the presumed site of injection. Foreign material was identified multifocally in the submucosa (Fig. 3A). The brain was liquefied, indicating hypoxic damage. Both lungs were congested and inflamed, with the right lung weighing 1,050 g and the left lung weighing 850 g.
Fig. 3.
Macroscopic and microscopic findings of the organs in case 2. (A) Filler was injected into the vaginal wall. Collagen filler (B) and a hybrid filler (C) consisting of hyaluronic acid and polylactic acid are observed in the vaginal wall, with vascular embolization (arrow in B) of the collagen filler (B and C, H&E, ×20). Embolism of the collagen filler in the lungs (D), heart (E), and brain (F) (D, H&E, ×40; E, H&E, ×100; F, H&E, ×200).
Microscopic examination of the vagina revealed two different types of fillers in the submucosal and muscular layers. Embolization of the collagen filler, but not the hybrid filler, was observed in some blood vessels (Fig. 3B, C). Collagen filler was also observed in many interstitial vessels of the lungs (Fig. 3D), suggesting that the initial cardiopulmonary arrest was caused by NTPE. Furthermore, emboli were observed in the heart and brain, indicating progression to a systemic embolism (Fig. 3E, F).
This study was approved by the National Forensic Service Institutional Review Board (906-240625-BR-001-01).
Discussion
NTPE occurs when substances that are not present in the blood vessels in significant amounts, such as fat, amniotic fluid, or air, are embolized into the pulmonary circulation [15]. Although rare, NTPE can be caused by filler injections if the filler is directly injected into veins or migrates into veins owing to high local pressure [4,5]. Similar cases of NTPE occurring after vaginal filler injections have been reported, with more than half of them resulting in death (Table 1). Most cases were caused by the injection of silicone oil, which was once popularly used. Reports on NTPE following the injection of other substances are relatively few, with only two autopsy cases reported [16,17]. This is the first autopsy case report in Korea. Owing to the significant risk of vascular complications after vaginal filler injections, a complete autopsy, including microscopic examination of the organs, should be performed. Clinicians should also be aware of these risks and refrain from performing such procedures.
Table 1.
Cases of non-thrombotic pulmonary embolism after vaginal filler injections
| Study | Age (yr) | Site | Material, dose (mL) | Symptom | Onseta) | Imaging | Pathology | Lab | Unlicensed practice | Outcomeb) |
|---|---|---|---|---|---|---|---|---|---|---|
| Jung et al. (1993) [9] | 43 | Vagina | Silicone, - | Dyspnea, BTS | 2 days | CT: bilateral consolidation; V/Q scan: reduced peripheral perfusion | Silicone emboli | WBC ↑; PaO2 ↓ | Yes | Recovery |
| 33 | Vagina | Silicone, - | Dyspnea, chest pain | Few hours | X-ray: bilateral diffuse infiltration | Silicone emboli, hemorrhage | WBC ↑; PaO2 ↓ | Yes | Death after 3 days | |
| Kang et al. (1999) [10] | 39 | Vagina | Silicone, 30 | Dyspnea | Immediate | CT: diffuse GGO and consolidation; V/Q scan: multiple perfusion defects | - | WBC ↑; PaO2 ↓ | Yes | Recovery |
| Seo et al. (1999) [11] | 43 | Vagina, breast | Silicone,30 (vagina),40 (breast) | Dyspnea, cough | 2 hr | CT: bilateral peripheral consolidation and GGO | Silicone emboli | WBC, D-dimer ↑; IgE WNL; PaO2 ↓ | Yes | Death after 1 wk |
| 38 | Vagina, breast | Silicone,20 (vagina),50 (breast) | BTS, dyspnea, cough | 2 hr | CT: bilateral peripheral consolidation and GGO; V/Q scan: peripheral perfusion defects | Silicone emboli | WBC ↑; D-dimer, IgE WNL; PaO2 ↓ | Yes | Recovery | |
| Chung et al.(2002) [12] | 44 | Vagina, breast, shoulder | Silicone, - | Chest pain | 6 hr | X-ray: bilateral diffuse haziness | Silicone emboli | - | Yes | Death after 15 hr |
| 39 | Vagina | Silicone, - | Chest pain, LOC | 2 hr | - | Silicone emboli | - | Yes | Death after 10 hr | |
| 32 | Vagina | Silicone, - | LOC | <1 hr | - | Silicone emboli, hemorrhage | - | No | Death after 5 days | |
| 58 | Vagina, buttock | Silicone, - | Fever, BTS, dyspnea | 2 days | - | Silicone emboli, acute pneumonitis, hemorrhage | - | Yes | Recovery | |
| 46 | Vagina | Silicone, - | Dyspnea, paralysis | 2 days | - | Silicone emboli, DAD | - | No | Death after 1 mo | |
| Kim et al. (2003) [13] | 46 | Vagina | Silicone, 10 | Dyspnea, chest pain | 2 days | CT: bilateral diffuse consolidation; V/Q scan: negative | Vacuoles in alveolar macrophages | PaO2 ↓ | Yes | Recovery |
| Park et al. (2010) [7] | 49 | Vagina | HA, 5 | Dyspnea, cough | 0 day | CT: bilateral diffuse GGO | Basophilic emboli | WBC, CRP ↑; PaO2 ↓ | Yes | Recovery |
| Duan et al. (2014) [16] | 34 | Vagina | PAAG, - | Abdominal pain, N/V, diarrhea | 1 day | - | Basophilic emboli, focal hemorrhages | - | Yes | Death after 1 day |
| Han et al. (2019) [18] c) | 56 | Vagina | HA, 15 | Hemoptysis, dyspnea | 1 day | CT: diffuse GGO and consolidation without embolism or RVE | - | CRP ↑; WBC, PLT, PT(INR), D-dimer WNL; Hb ↓ | Yes | Recovery |
| Yang et al. (2020) [8] | 33 | Vagina | HA, 13; collagen, 2 | Dyspnea | 5-10 min | CT: RAE and RVE with PA filling defects | - | WBC, CRP, D-dimer ↑ | No | Death after 9 hr |
| Kong et al. (2023) [17] | 40 | Vagina | HA, - | Abdominal pain | 2 hr | CT: GGO in both lower lobes | Edema, amorphous basophilic emboli in interstitial vessels | WBC, PT(INR), aPTT ↑; IgE WNL; PLT ↓ | Yes | Death after 19 hr |
| Present case | 38 | Vagina | HA, 15(62 in total) | LOC | 20-40 min | CT: RVE and dilation of pulmonary trunk with no PA filling defects | Basophilic emboli, hemorrhage, DAD | D-dimer, CK-MB, troponin T ↑ | No | Death after 10 days |
| 35 | Vagina | Collagen, -; HA+PLA, - | Desaturation, cardiac arrest | 4 min | - | Collagen emboli, pneumonitis | - | No | Death after 1 mo |
BTS, blood-tinged sputum; CT, computed tomography; V/Q, ventilation/perfusion; WBC, white blood cell; GGO, ground-glass opacity; WNL, within normal limits; LOC, loss of consciousness; DAD, diffuse alveolar damage; HA, hyaluronic acid; CRP, C-reactive protein; PAAG, polyacrylamide hydrogel; N/V, nausea and vomiting; RVE, right ventricular enlargement; PLT, platelet; PT, prothrombin time; INR, international normalized ratio; RAE, right atrial enlargement; aPTT, activated partial thromboplastin time; PA, pulmonary artery; CK-MB, creatine kinase– myocardial band; PLA, polylactic acid.
NTPE can also result from filler injections into the face, breasts, buttocks, and joints [14,19-21]. However, according to a study of silicone embolism syndrome by Schmid et al. [14], embolisms after vaginal injections were more often fatal and frequently accompanied by neurological changes. The vagina is surrounded by an extensive venous plexus, which makes it a dangerous site for filler injections. Even when doctors take precautions such as aspirating the syringe before injection, there remains a risk of direct intravascular injection. Additionally, because the vagina is located in a confined space, the pressure can easily increase when a large amount of filler is injected or when injected rapidly. In both cases, a large amount of filler was detected at autopsy, with grossly visible embolization in case 1. The large injection volume may have caused the filler to migrate into the paravaginal veins, resulting in NTPE.
A literature review revealed that 16 cases of NTPE after vaginal filler injections have been reported (Table 1). The injected materials were silicone (n=11), HA (n=3), a combination of HA and collagen (n=1), or polyacrylamide hydrogel (PAAG) (n=1), with nine cases resulting in death. The time span from the procedure to symptom onset was as short as 5–10 minutes, with some patients even developing symptoms during injection [14]. Our patients showed rapid progression of symptoms, initially leading to the suspicion of anaphylaxis. However, in case 1, the symptoms were not alleviated by epinephrine, suggesting that NTPE was the cause of the symptoms rather than a hypersensitivity reaction. NTPE was pathologically confirmed in all examined cases, even in those without evidence from imaging studies (case 1 and a case reported by Kim et al. [13]). Intra-alveolar hemorrhage was also a common pathological finding regardless of the injected material, possibly resulting from ischemic damage to the lungs. Emboli in vaginal vessels were also detected in autopsied cases [11,16,17]. The overall pathological findings were similar to our cases.
Forensic pathologists should consider NTPE as a potential cause of death after vaginal filler injections and perform a complete autopsy. A thorough review of EMS and hospital records, including laboratory tests and imaging studies, is mandatory. The possibility of NTPE persists even when imaging studies are negative; therefore, microscopic examination of the lungs should be performed. Additionally, as in case 2, there may be no noticeable gross abnormalities, even when a systemic embolism has occurred. Therefore, microscopic examination of other major organs, including the heart and brain, should be performed. Special stains can help diagnose embolisms: HA and PAAG stain positive with Alcian blue [16,17] and collagen stains positive with Masson's trichrome staining. Gross examination after formalin fixation is also helpful because many fillers are gelatinous. Furthermore, NTPE can be reported as an adverse device effect to the National Institute of Food and Drug Safety Evaluation (NIFDS) to raise awareness and help inform preventive measures.
One purpose of the death investigation system is to prevent premature deaths. We believe that measures are needed to enhance patient safety and prevent deaths resulting from this potentially life-threatening complication of medically unnecessary procedures. Many OB-GYN societies worldwide, including those in the United States, Canada, the United Kingdom, Australia, and New Zealand, are currently opposed to performing female genital cosmetic procedures [22-25]. In Korea, the intended use of fillers approved by the Medical Device Evaluation Department at the NIFDS includes the improvement of facial wrinkles, penile enlargement, and volume restoration of the face or hands [6]. Vaginal injections are not approved or indicated. As recommended by the OB-GYN societies, doctors should first explain to patients that the vaginal anatomy varies from person to person and that the efficacy and safety of the procedure have not been proven. Even if patients request filler injections despite the explanation, doctors should avoid performing the procedure or at least fully inform patients about this fatal complication.
Finally, regulations are required to prevent illegal procedures. Vaginal filler injections were performed by unlicensed practitioners in 81% (13/16) of the reported cases (Table 1). These practitioners may be unaware of the rich venous plexus surrounding the vagina and may not follow basic procedural rules, such as aspiration before injection. Furthermore, they may inject a large volume simultaneously or apply excessive pressure during injection, leading to more complications. This poses a significant threat to patient safety.
In conclusion, this report highlighted NTPE as a rare but fatal complication of vaginal filler injections. Forensic pathologists should keep this in mind and perform thorough examinations during autopsies. Moreover, clinicians must recognize the risks associated with the procedures performed and adhere to the approved use of fillers.
REFERENCES
1.American Society of Plastic Surgeons. 2023 ASPS procedural statistics release [Internet]. Arlington Heights, IL: American Society of Plastic Surgeons; 2023. [cited 2024 Jul 14]. Available from: https://www.plasticsurgery.org/documents/news/statistics/2023/plastic-surgery-statistics-report-2023.pdf.
2.Vanaman M, Bolton J, Placik O, et al. Emerging trends in nonsurgical female genital rejuvenation. Dermatol Surg 2016;42:1019–29.
[Article] [PubMed]
[Article] [PubMed]
3.Ortiz AE, Ahluwalia J, Song SS, et al. Analysis of U.S. Food and Drug Administration data on soft-tissue filler complications. Dermatol Surg 2020;46:958–61.
[Article] [PubMed]
[Article] [PubMed]
4.DeLorenzi C. Complications of injectable fillers, part 2: vascular complications. Aesthet Surg J 2014;34:584–600.
[Article] [PubMed]
[Article] [PubMed]
5.Soares DM. Bridging a century-old problem: the pathophysiology and molecular mechanisms of HA filler-induced vascular occlusion (FIVO): implications for therapeutic interventions. Molecules 2022;27:5398.
[Article] [PubMed] [PMC]
[Article] [PubMed] [PMC]
6.Guideline on the review and approval of cosmetic fillers [Internet]. Cheongju: Medical Device Evaluation Department, National Institute of Food and Drug Safety Evaluation, Ministry of Food and Drug Safety; 2020. [cited 2024 Jun 12]. Available from: https://www.mfds.go.kr/brd/m_1060/view.do?seq=14677&srchFr=&srchTo=&srchWord=&srchTp=&itm_seq_1=0&itm_seq_2=0&multi_itm_seq=0&company_cd=&company_nm=&page=5.
7.Park HJ, Jung KH, Kim SY, et al. Hyaluronic acid pulmonary embolism: a critical consequence of an illegal cosmetic vaginal procedure. Thorax 2010;65:360–1.
[Article] [PubMed]
[Article] [PubMed]
8.Yang Y, Sheng H, Gu Q, et al. Death caused by vaginal injection of hyaluronic acid and collagen: a case report. Aesthet Surg J 2020;40:NP263–8.
[Article] [PubMed]
[Article] [PubMed]
9.Jung BH, Suh YI, Lee JM, et al. Two cases of silicone-induced pulmonary embolism. Tuberc Respir Dis 1993;40:610–5.
[Article]
[Article]
10.Kang MB, Kim ST, Lee JG, et al. A case of acute respiratory distress syndrome with pulmonary embolism induced by injection of silicone at vaginal wall. Tuberc Respir Dis 1999;46:414–9.
[Article]
[Article]
11.Seo YB, Chung JH, Park SM, et al. Four case of silicone fluid induced acute pulmonary embolism. Korean J Med 1999;57:364–75.
12.Chung KY, Kim SH, Kwon IH, et al. Clinicopathologic review of pulmonary silicone embolism with special emphasis on the resultant histologic diversity in the lung: a review of five cases. Yonsei Med J 2002;43:152–9.
[Article] [PubMed]
[Article] [PubMed]
13.Kim CH, Chung DH, Yoo CG, et al. A case of acute pneumonitis induced by injection of silicone for colpoplasty. Respiration 2003;70:104–6.
[Article] [PubMed]
[Article] [PubMed]
14.Schmid A, Tzur A, Leshko L, et al. Silicone embolism syndrome: a case report, review of the literature, and comparison with fat embolism syndrome. Chest 2005;127:2276–81.
[Article] [PubMed]
[Article] [PubMed]
15.Jorens PG, Van Marck E, Snoeckx A, et al. Non-thrombotic pulmonary embolism. Eur Respir J 2009;34:452–74.
[Article] [PubMed]
[Article] [PubMed]
16.Duan Y, Zhang L, Li S, et al. Polyacrylamide hydrogel pulmonary embolism. A fatal consequence of an illegal cosmetic vaginal tightening procedure: a case report. Forensic Sci Int 2014;238:e6–10.
[Article] [PubMed]
[Article] [PubMed]
17.Kong J, Yang T, Yang X, et al. Death from pulmonary embolism caused by vaginal injection of hyaluronic acid: a case report and a literature review. Aesthetic Plast Surg 2023;47:1535–41.
[Article] [PubMed]
[Article] [PubMed]
18.Han SW, Park MJ, Lee SH. Hyaluronic acid-induced diffuse alveolar hemorrhage: unknown complication induced by a well-known injectable agent. Ann Transl Med 2019;7:13.
[Article] [PubMed] [PMC]
[Article] [PubMed] [PMC]
19.Bhagat R, Forteza RM, Calcote CB, et al. Pulmonary emboli from therapeutic sodium hyaluronate. Respir Care 2012;57:1670–3.
[Article] [PubMed]
[Article] [PubMed]
20.Jang JG, Hong KS, Choi EY. A case of non-thrombotic pulmonary embolism after facial injection of hyaluronic Acid in an illegal cosmetic procedure. Tuberc Respir Dis (Seoul) 2014;77:90–3.
[Article] [PubMed] [PMC]
[Article] [PubMed] [PMC]
21.Inayat F, Cheema AR, Virk HU, et al. Non-thrombotic pulmonary embolism: a potential complication of polyacrylamide hydrogel cosmetic injection. Case Rep Med 2016;2016:1397434.
[Article] [PubMed] [PMC]
[Article] [PubMed] [PMC]
22.Ethical considerations in relation to female genital cosmetic surgery (FGCS) [Internet]. London: Royal College of Obstetricians & Gynaecologists; 2013. [cited 2024 Jul 14]. Available from: https://qna.files.parliament.uk/qna-attachments/788291/original/HL3083%20B.pdf.
23.RANZCOG Women's Health Committee. Vaginal ‘rejuvenation’ and cosmetic vaginal procedures. Aust N Z J Obstet Gynaecol 2019;59:E22–3.
[Article] [PubMed]
[Article] [PubMed]
24.Elective female genital cosmetic surgery: ACOG Committee opinion, number 795. Obstet Gynecol 2020;135:e36–42.
[PubMed]
[PubMed]