Figures
Abstract
Renal failure is a common morbidity in multiple myeloma (MM). Although proteinuria has been increasingly reported in malignancies, it is not routinely used to refine risk estimates of survival outcomes in patients with MM. Here we aimed to investigate initial serum albumin and 24-hour daily protein excretion (24-h DPE) before treatment as prognostic factors in patients with MM. We conducted a retrospective analysis of 102 patients with myeloma who were ineligible for haematopoietic stem cell transplantation between October 2000 and December 2012. Initial proteinuria was assessed before treatment by quantitative analysis of 24-hour urine samples. The demographic and laboratory characteristics, survival outcome, and significance of pre-treatment 24-h DPE and albumin in the new staging system of MM were analyzed. Pre-treatment proteinuria (>300 mg/day) was present in 66 patients (64.7%). The optimal cut-off value of 24-h DPE before treatment was 500 mg/day. Analysis of the time-dependent area under the curve showed that the serum albumin and 24-h DPE before treatment were better than 24-h creatinine clearance rate and β2-microglobulin. A subgroup analysis showed that an initial excess proteinuria (24-h DPE ≥ 500 mg) was associated with poor survival status (17.51 vs. 34.24 months, p = 0.002). Furthermore, initial serum albumin was an independent risk factor on multivariate analysis (<2.8 vs. ≥2.8, hazard ratio = 0.486, p = 0.029). Using the A-DPE staging system, there was a significant survival difference among patients with stage I, II, and III MM (p < 0.001). Initial serum albumin and 24-h DPE before treatment showed significant prognostic factors in patients with MM, and the new A-DPE staging system may be utilized instead of the International Staging System. Its efficacy should be evaluated by further large prospective studies.
Citation: Chen J-H, Hsu S-N, Huang T-C, Wu Y-Y, Lin C, Chang P-Y, et al. (2015) Prognostic Significance of Initial Serum Albumin and 24 Hour Daily Protein Excretion before Treatment in Multiple Myeloma. PLoS ONE 10(6): e0128905. https://doi.org/10.1371/journal.pone.0128905
Received: January 31, 2015; Accepted: May 2, 2015; Published: June 8, 2015
Copyright: © 2015 Chen et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited
Data Availability: All relevant data are within the paper.
Funding: The authors have no support or funding to report.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Multiple myeloma (MM) is a neoplastic disorder characterized by a single clone abundance of plasma cells occupying in the bone marrow and generating a monoclonal immunoglobulin, which sequentially results in end organ damage and related complications such as anemia, renal insufficiency, hypercalcaemia, skeletal events, and infection [1–4]. Although proteinuria is not uncommon in patients with malignancies, a high prevalence of proteinuria was reported in patients with MM [5–7]. Furthermore, the correlation between proteinuria and MM is well established, but its clinical impact on survival has not yet been elucidated owing to a lack of clinical studies.
Renal involvement in MM can manifest as subclinical proteinuria to overt proteinuria or even nephropathy. Proteinuria-induced renal failure remains a major cause of morbidity and mortality in patients with MM [8–10]. The earlier the magnitude of proteinuria is reduced, the lower the risks of renal disease progression and mortality become [11, 12]. Moreover, with the introduction of bortezomib-based treatment, survival has improved greatly in recent years when early improvement in renal function occurred [13–15]. The Durie Salmon (DS) staging system, which was primarily used 40 years previously for patients with MM, was designed with each stage divided into A and B subgroups according to renal function [16]. The International Staging System (ISS) published by the International Myeloma Working Group in 2005 introduced a new staging system using β2-microglobulin (β2M) and albumin levels as prognostic factors [17]. Whether the DS or ISS is used, renal function and albumin have been considered easy and good indicators of survival [18]. However, β2M is easily influenced by many factors including renal function, diverse autoimmune disease, and haematological malignancies [19–23]. Therefore, we try to introduce new parameters, including 24-h daily protein excretion (DPE) and albumin, to refine risk estimates of survival outcomes in patients with MM.
In this study, we investigated the initial serum albumin and 24-h DPE before treatment as prognostic factors in patients with MM who were ineligible to undergo haematopoietic stem cell transplantation (HSCT) and determined the significance of the new staging system of MM. We also evaluated the true incidence and clinical impact of proteinuria and other confounding factors on patients with MM.
Patients and Methods
Study population and diagnosis of MM
This retrospective analysis enrolled a total of 102 patients with MM who were ineligible for HSCT in the Tri-Service General Hospital (TSGH) between October 2000 and December 2012 (Fig 1). Because the patient records/information were anonymized and de-identified prior to analysis in this study, informed consent was not required. The study was performed under the guidelines of the Helsinki Declaration and approved by the Human Subjects Protection Offices (institutional review board) of TSGH, National Defense Medical Center in Taiwan. All patients had symptomatic MM in accordance with the diagnostic criteria of the International Myeloma Working Group. All enrolled patients received treatment after the collection of 24-h urine samples for the measurement of creatinine clearance rate (Ccr) and DPE. The clinical information collected from the medical records included: age, sex, Eastern Cooperative Oncology Group performance status (ECOG PS) at diagnosis, disease severity, survival duration, and pre-treatment laboratory data, such as serum albumin, blood urea nitrogen (BUN), creatinine (Cr), total calcium, β2M, urinary 24-h Ccr, and 24-h DPE. The disease severity at diagnosis was defined by a different prognostic system [24, 25]. In addition, the impact of variable confounding factors on overall survival (OS) was analyzed in this study population.
Assessment of initial albumin before treatment
Pre-treatment serum albumin was assessed by quantitative analysis of serum samples (g/dL). The demographic and laboratory characteristics, prevalence, survival outcome, and significance of pre-treatment serum albumin in the MM staging system were analyzed.
Assessment of initial proteinuria before treatment
Proteinuria was defined as excreting > 300 mg of total protein per 24 hours (24-h DPE > 300 mg) according to the Kidney Disease Outcomes Quality Initiative guidelines [26]. Quantitative analysis data of the 24-h urine samples before treatment were analyzed to ascertain the DPE and Ccr values. Urinary Cr and protein concentrations were determined using Jaffe and colorimetric methods, respectively. The total urinary protein (mg/day) was determined by multiplying the total urine volume (dL) by the protein concentration in the test sample (mg/dL).
MM treatment protocol
Patients with MM who were ineligible for the HSCT received various treatments; they could be divided into three groups: those who received conservative treatment (no treatment or dexamethasone 20 mg/m2 orally only), those who received conventional chemotherapy and thalidomide (e.g., MP, VAD, TD, and TMD), and those who received novel agents with bortezomib-based treatment (BTD).
The MP regimen consisted of melphalan (8 mg/m2) and prednisolone (60 mg/m2) for 4 days; the VAD regimen consisted of vincristine (0.4 mg) and doxorubicin (9 mg/m2) for 4 days and dexamethasone (40 mg) for 12 days; and the TD regimen consisted of 200 mg thalidomide orally per day plus 5–10 mg dexamethasone intravenously 4 times per day. The BTD arm was identical to the TD arm plus bortezomib 1.3 mg/m2 on days 1, 4, 8, and 11 of each cycle.
Statistical analysis
Data were analyzed using SPSS. In addition to descriptive and frequency statistics, univariate and multivariate analyses were used to evaluate the significance of all confounding factors in the patients with MM. To further evaluate the predictive ability of initial serum albumin, β2M, 24-h Ccr, and 24-h DPE, we conducted a time-dependent receiver operating characteristic curve (ROC) analysis for censored records. This analysis uses sensitivity and specificity, both of which are time-dependent, to measure the prognostic capacity of a survival model as measured by the area under the curve (AUC) [27]. Compared with survival status, the Kaplan-Meier curve was used to present their differences and the log-rank test was used to analyze the data. We considered p-values < 0.05 as significant for all analyses.
Results
Patient characteristics
Since October 2000, a total of 102 patients (median age, 71.0 years; median OS, 18.0 months) with MM were recruited in this study. The median survival time of stage I, II, III by ISS staging system were 67, 25 and 15 months (P = 0.049). On average, myeloma affects male patients more often than female patients at a ratio of 1.61:1. Table 1 summarizes the patients’ baseline characteristics, myeloma types, and survival duration. A total of 56 patients (54.9%) were diagnosed with the immunoglobulin G (IgG) type myeloma, followed by the light chain type (19.6%) and immunoglobulin A (IgA) type (18.6%). Only 7 patients (6.9%) had the non-secretory type of myeloma. However, 27 patients (26.5%) had renal function impairment (serum Cr ≥ 2.0 mg/dL), among which 12 (44%) were associated with light chain type myeloma, 8 (30%) with IgG type, 6 (22%) with IgA type, and 1 (4%) with non-secretory type. Sixty-six of the 102 patients (64.7%) had proteinuria prior to treatment. A total of 40 patients (39.2%) received BTD in our study population.
Time-dependent ROC analysis and the impact of proteinuria
In our study, the parameters associated with renal function include serum β2M, albumin, 24-h Ccr, and 24-h DPE. The optimal cut-off value of the initial 24-h DPE was 500 mg/day on ROC analysis. According to time-dependent AUC analysis (Fig 2), the prognostic capacity of initial serum albumin and 24-h DPE before treatment was better than those of 24-h Ccr and β2M. In addition, the pre-treatment albumin level played the most important role in survival capacity, especially within 22 months after anti-myeloma treatment.
On subgroup univariate analysis using the optimal cut-off value of proteinuria (500 mg/day), patients with initial excess proteinuria (≥500 mg/day) had a significantly poorer ECOG PS (p = 0.018), higher serum BUN (p = 0.004), higher serum Cr (p = 0.020), lower serum albumin levels (p = 0.009), and shorter survival duration (17.51 vs. 34.24 months, p = 0.002) (Table 2).
Analysis of all confounding factors on OS
Subsequent analysis of all variables associated with all-cause mortality was performed by univariate and multivariate analyses. The univariate analysis of the continuous method showed that age, ECOG PS, ISS stage, proteinuria, 24-h Ccr, serum BUN, total calcium, and albumin were significant prognostic factors (Table 3). Using the dichotomous method, the results revealed that ECOG PS, stage ISS, proteinuria, 24-h Ccr, serum BUN, Cr, total calcium, albumin, and the effect of bortezomib were significant. However, serum β2M was not significant on univariate analysis using both the continuous and dichotomous methods.
Further multivariate analysis of the prognostic factors using the Cox proportional hazard model showed that only serum BUN and Cr were significant prognostic factors on the continuous method. Using the dichotomous method, however, the result revealed that pre-treatment serum albumin (<2.8 mg/dL) was an independent prognostic factor (<2.8 vs. ≥2.8; hazard ratio, 0.486; 95% confidence interval, 0.254–0.930; p = 0.029). Similarly, serum β2M was not significant in multivariate analysis by using both the continuous and dichotomous methods (Table 4).
A-DPE staging system
According to the results of time-dependent AUC, multivariate, and univariate analyses, we formulated a new experimental staging system, the A-DPE staging system, based on initial serum albumin and 24-h DPE prior to treatment. In this system, stage I defined as albumin ≥ 2.8 g/dL and 24-h DPE < 500 mg/dL; stage II as albumin ≥ 2.8 g/dL and 24-h DPE ≥ 500 mg/dL; and stage III as albumin < 2.8 g/dL. Applying this system, we found that 37 patients had stage I, 38 had stage II, and 27 had stage III. According to this staging system, there was a significant survival difference among myeloma patients with stage I, II, and III disease (p < 0.001, Fig 3).
Discussion
In patients with MM, the significance of initial serum albumin and 24-h DPE before treatment led to the introduction of the new A-DPE staging system, and we found a significant survival difference among the different stages (p < 0.001). In addition, our results indicate that 64.7% (66/102) of the patients had proteinuria, which was not uncommon as previous studies revealed. It also demonstrated that patients with excess proteinuria (24-h DPE ≥ 500 mg/day) tended to have poor survival status (17.51 vs. 34.24 months, p = 0.002) and that those with lower albumin levels (<2.8 g/dL, A-DPE stage III) had the worst survival rate (p < 0.001). To date, our study was the first to investigate the impact and prevalence of proteinuria and significance of 24-h DPE in patients with MM. These findings are important because the current clinical practice for predicting outcomes in patients with MM are solely based on serum albumin and β2M without explicit consideration of the severity of concomitant proteinuria.
Proteinuria, a marker of kidney disease, is strongly associated with a risk of adverse outcomes [28–30]. Many studies have demonstrated that the prevalence of proteinuria increased with poor control of blood pressure, a high glycated haemoglobin level, and accompanying renal or malignant disease [6, 29, 31]. In addition, patients with malignancies often present with proteinuria but without significant symptoms to emphasize its clinical importance. Thus, we use quantitative analysis of 24-h urine to ascertain DPE and renal function. Despite its evolving role as a major risk factor of all-cause mortality, little is known about the clinical impact of proteinuria in patients with MM. In an attempt to choose a more significant prognostic factor and determine the correlation between 24-h DPE and kidney function, we were unable to find any study that included 24-h DPE as a prognostic factor. Thus, to our knowledge, this study is the first to investigate serum albumin and 24-h DPE as prognostic factors in patients with MM. Furthermore, the parameters associated with renal function include not only albumin and 24-h DPE but also serum β2M, and 24-h Ccr. Instead of β2M, our new staging system used 24-h DPE owing to β2M not being significant if kidney function was relatively normal. Of note, our study proved that serum β2M was not significant on univariate or multivariate analysis using both the continuous and dichotomous methods. Thus, the effect of β2M on survival outcomes might be mainly owing to declining kidney function.
On time-dependent AUC analysis, the results indicate that 24-h DPE is a new potential prognostic factor in patients with MM, especially for long-term survival outcomes. Moreover, it demonstrates that pre-treatment serum albumin has the better predictive value in the early course of the disease, whereas pre-treatment 24-h DPE has better predictive value in the late course of the disease. In addition, multivariate analysis of the prognostic factors revealed that pre-treatment serum albumin was the most important independent risk factor of all-cause mortality. Thus, we incorporate pre-treatment serum albumin and 24-h DPE into a discrete new staging system for predicting outcomes in patients with MM. Survival curves of different stages were discrete; they were able to show that 24-h DPE was a more significant and powerful prognostic factor than β2M.
Recent studies showed that bortezomib appears to be most valuable in this setting of treatment with MM and renal impairment and that, when combined with other drugs, increases the probability of rapid remission and related improvements in renal function [32]. Substantial evidence indicates that the administration of bortezomib and high-dose dexamethasone with or without a third drug (e.g., cyclophosphamide, thalidomide, or doxorubicin) is an appropriate option for patients with any degree of renal impairment [15, 33]. In the future, large, prospective cohort studies may be needed to confirm whether the bortezomib-based regimen could overcome poor prognosis in the presence of any degree of renal impairment or proteinuria.
However, our study had some inherent limitations. First, because of its retrospective design and small sample size, patient heterogeneity, and different treatment protocols, the prognostic impact of proteinuria on patients with MM needs to be confirmed by large, prospective cohort studies. Second, the prevalence of proteinuria may be underestimated because we only enrolled patients for whom a 24-h DPE and chart were available may create some selection bias. Third, the majority of enrolled patients did not undergo renal biopsy to clarify whether the proteinuria was caused solely by paraneoplastic syndrome or other systemic disorders that involve the kidneys. Fourth, the information of genetic abnormalities was lacking because of the lack of a routine to perform cytogenetic studies and equipment to perform fluorescent in situ hybridisation in our institute. Therefore, it was difficult to identify who had high-risk genetic factors, which may affect the choice of frontline treatment and impair survival outcomes. Fifth, few patients received newer therapeutic strategies such as frontline bortezomib–containing regimens. Only 3 patients received bortezomib as the first line treatment since June 2012. Finally, the follow-up duration was too short to evaluate the survival benefit of treatment.
In summary, here we demonstrated that the prevalence of proteinuria is not uncommon in patients with MM. In addition, we found that the initial 24-h DPE combined with serum albumin before treatment possess powerful and significant prognostic value for predicting survival outcomes in patients with MM. Routine screening of pre-treatment proteinuria in patients with MM may help clinicians provide better patient care. Whether reducing proteinuria improves survival requires further investigations.
Supporting Information
S1 Fig. According to the calculation of AUC by time, and added the calculated figures.
https://doi.org/10.1371/journal.pone.0128905.s001
(PDF)
S2 Fig. According to the calculation of AUC by time, and added the calculated figures.
https://doi.org/10.1371/journal.pone.0128905.s002
(PDF)
S3 Fig. Overall survival of multiple myeloma based on clinical stage at diagnosis by ISS staging.
https://doi.org/10.1371/journal.pone.0128905.s003
(TIF)
Author Contributions
Conceived and designed the experiments: JHC SNH. Performed the experiments: JHC CLH. Analyzed the data: CL YYW. Contributed reagents/materials/analysis tools: TCH YYW PYC YCC CLH. Wrote the paper: JHC SNH.
References
- 1. Alexanian R, Dimopoulos M (1994) The treatment of multiple myeloma. The New England journal of medicine. 330(7):484–9. pmid:8289856
- 2. Palumbo A, Anderson K (2011) Multiple myeloma. The New England journal of medicine. 364(11):1046–60. pmid:21410373
- 3. Kyle RA, Gertz MA, Witzig TE, Lust JA, Lacy MQ, Dispenzieri A, et al. (2003) Review of 1027 patients with newly diagnosed multiple myeloma. Mayo Clinic proceedings. 78(1):21–33. pmid:12528874
- 4. Prakash J, Niwas SS, Parekh A, Vohra R, Wani IA, Sharma N, et al. (2009) Multiple myeloma—presenting as acute kidney injury. The Journal of the Association of Physicians of India. 57:23–6. pmid:19753754
- 5. Yun JP, Suh C, Lee E, Chang JW, Yang WS, Park JS, et al. (2006) Comparison of serum beta 2-microglobulin and 24 hour urinary creatinine clearance as a prognostic factor in multiple myeloma. Journal of Korean medical science. 21(4):639–44. pmid:16891806
- 6. Sawyer N, Wadsworth J, Wijnen M, Gabriel R (1988) Prevalence, concentration, and prognostic importance of proteinuria in patients with malignancies. British medical journal. 296(6632):1295–8. pmid:3133055
- 7. Norris SH (1993) Paraneoplastic glomerulopathies. Seminars in nephrology. 13(3):258–72. pmid:8321926
- 8. Hutchison CA, Batuman V, Behrens J, Bridoux F, Sirac C, Dispenzieri A, et al. (2012) The pathogenesis and diagnosis of acute kidney injury in multiple myeloma. Nature reviews Nephrology. 8(1):43–51.
- 9. Faiman BM, Mangan P, Spong J, Tariman JD, The International Myeloma Foundation Nurse Leadership B (2011) Renal complications in multiple myeloma and related disorders: survivorship care plan of the International Myeloma Foundation Nurse Leadership Board. Clinical journal of oncology nursing. 15 Suppl:66–76. pmid:21816711
- 10. Eleftherakis-Papapiakovou E, Kastritis E, Roussou M, Gkotzamanidou M, Grapsa I, Psimenou E, et al. (2011) Renal impairment is not an independent adverse prognostic factor in patients with multiple myeloma treated upfront with novel agent-based regimens. Leukemia & lymphoma. 52(12):2299–303.
- 11. Bakris GL (2008) Slowing nephropathy progression: focus on proteinuria reduction. Clinical journal of the American Society of Nephrology: CJASN. 3 Suppl 1:S3–10. pmid:18178794
- 12. Merchionne F, Procaccio P, Dammacco F (2008) Long-term survival in multiple myeloma: a single-center experience. Clinical and experimental medicine. 8(3):133–9. pmid:18791685
- 13. Dimopoulos MA, Roussou M, Gavriatopoulou M, Zagouri F, Migkou M, Matsouka C, et al. (2009) Reversibility of renal impairment in patients with multiple myeloma treated with bortezomib-based regimens: identification of predictive factors. Clinical lymphoma & myeloma. 9(4):302–6.
- 14. Kastritis E, Terpos E, Dimopoulos MA (2013) Current treatments for renal failure due to multiple myeloma. Expert opinion on pharmacotherapy. 14(11):1477–95. pmid:23705905
- 15. Gaballa MR, Laubach JP, Schlossman RL, Redman K, Noonan K, Mitsiades CS, et al. (2012) Management of myeloma-associated renal dysfunction in the era of novel therapies. Expert review of hematology. 5(1):51–66; quiz 7–8. pmid:22272706
- 16. Durie BG, Salmon SE (1975) A clinical staging system for multiple myeloma. Correlation of measured myeloma cell mass with presenting clinical features, response to treatment, and survival. Cancer. 36(3):842–54. pmid:1182674
- 17. Greipp PR, San Miguel J, Durie BG, Crowley JJ, Barlogie B, Blade J, et al. (2005) International staging system for multiple myeloma. Journal of clinical oncology: official journal of the American Society of Clinical Oncology. 23(15):3412–20. pmid:15809451
- 18. Jacobson JL, Hussein MA, Barlogie B, Durie BG, Crowley JJ (2003) A new staging system for multiple myeloma patients based on the Southwest Oncology Group (SWOG) experience. British journal of haematology. 122(3):441–50. pmid:12877671
- 19. Cuzick J, Cooper EH, MacLennan IC (1985) The prognostic value of serum beta 2 microglobulin compared with other presentation features in myelomatosis. British journal of cancer. 52(1):1–6. pmid:3893505
- 20. Alexanian R, Barlogie B, Fritsche H (1985) Beta 2 microglobulin in multiple myeloma. American journal of hematology. 20(4):345–51. pmid:3907332
- 21. Scarffe JH, Anderson H, Palmer MK, Crowther D (1983) Prognostic significance of pretreatment serum beta 2-microglobulin levels in multiple myeloma. European journal of cancer & clinical oncology. 19(10):1361–4.
- 22. Bataille R, Grenier J, Sany J (1984) Beta-2-microglobulin in myeloma: optimal use for staging, prognosis, and treatment—a prospective study of 160 patients. Blood. 63(2):468–76. pmid:6362753
- 23. Cuzick J, De Stavola BL, Cooper EH, Chapman C, MacLennan IC (1990) Long-term prognostic value of serum beta 2 microglobulin in myelomatosis. British journal of haematology. 75(4):506–10. pmid:2119793
- 24. Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D (1999) A more accurate method to estimate glomerular filtration rate from serum creatinine: a new prediction equation. Modification of Diet in Renal Disease Study Group. Annals of internal medicine. 130(6):461–70. pmid:10075613
- 25. Durie BG, Harousseau JL, Miguel JS, Blade J, Barlogie B, Anderson K, et al. (2006) International uniform response criteria for multiple myeloma. Leukemia. 20(9):1467–73. pmid:16855634
- 26. (2002) K/DOQI clinical practice guidelines for chronic kidney disease: evaluation, classification, and stratification. American journal of kidney diseases: the official journal of the National Kidney Foundation. 39(2 Suppl 1):S1–266.
- 27. Heagerty PJ, Lumley T, Pepe MS (2000) Time-dependent ROC curves for censored survival data and a diagnostic marker. Biometrics. 56(2):337–44. pmid:10877287
- 28. Turin TC, Tonelli M, Manns BJ, Ahmed SB, Ravani P, James M, et al. (2013) Proteinuria and life expectancy. American journal of kidney diseases: the official journal of the National Kidney Foundation. 61(4):646–8. pmid:23228947
- 29. Konta T (2013) Clinical significance of proteinuria and renal function: findings from a population-based cohort, the Takahata study. Rinsho byori The Japanese journal of clinical pathology. 61(7):629–34. pmid:24205706
- 30. Cravedi P, Ruggenenti P, Remuzzi G (2012) Proteinuria should be used as a surrogate in CKD. Nature reviews Nephrology. 8(5):301–6. pmid:22391456
- 31. Puolijoki H, Mustonen J, Pettersson E, Pasternack A, Lahdensuo A (1989) Proteinuria and haematuria are frequently present in patients with lung cancer. Nephrology, dialysis, transplantation: official publication of the European Dialysis and Transplant Association—European Renal Association. 4(11):947–50.
- 32.
Grzasko N, Morawska M, Hus M (2014) Optimizing the Treatment of Patients With Multiple Myeloma and Renal Impairment. Clinical lymphoma, myeloma & leukemia. Clin Lymphoma Myeloma Leuk. 2014 Oct 5. pii: S2152-2650(14)00422-4.
- 33. Chanan-Khan AA, San Miguel JF, Jagannath S, Ludwig H, Dimopoulos MA (2012) Novel therapeutic agents for the management of patients with multiple myeloma and renal impairment. Clinical cancer research: an official journal of the American Association for Cancer Research. 18(8):2145–63. pmid:22328563