The Impact of Lipoprotein apheresis on Plasma Protein Levels: A Meta‐Analysis

ABSTRACT

Lipoprotein apheresis (LA) is used for severe dyslipidemia but can remove non‐lipid plasma constituents. In this systematic review and meta‐analysis, we synthesized evidence on LA‐associated changes in albumin, total protein (TP), and α2‐macroglobulin (α2M). PubMed, Scopus, and Web of Science were searched from inception to August 2025 for human observational studies reporting pre‐/post‐apheresis values for albumin, TP, or α2M. Pooled effects were summarized as weighted mean differences (WMD) with 95% CIs. Prespecified subgroup analyses included apheresis duration (≥ 1 year vs. < 1 year), disease background, and apheresis method. Twenty‐eight publications met eligibility (albumin: 22 publications/39 studies; TP: 21 publications/34 studies; α2M: 6 publications/7 studies). LA was associated with lower post‐procedure levels of albumin (WMD = −3.80 g/L; 95% CI: −4.61 to −2.99; p < 0.001), TP (WMD = −8.73 g/L; 95% CI: −11.06 to −6.41; p < 0.001), and α2M (WMD = −65.79 mg/dL; 95% CI: −88.98 to −42.60; p < 0.001). Reductions were observed across durations (< 1 year and ≥ 1 year), disease groups, and apheresis methods, with larger albumin decreases in DALI‐based protocols and smaller in other methods. Between‐study heterogeneity was high but diminished in several subgroups; meta‐regression suggested background disease as a potential contributor. Across observational studies, LA reduced albumin, TP, and α2M immediately post‐procedure. Magnitudes vary by method, clinical context, and treatment duration. These findings support routine monitoring of protein profiles during maintenance LA.

1. Introduction

Lipoprotein apheresis (LA) remains an important option for patients with severe, treatment‐refractory dyslipidemia, including heterozygous or homozygous familial hypercholesterolemia and lipoprotein(a)‐driven disease, despite substantial advances in pharmacotherapy [1, 2]. While LA's therapeutic target is the removal of atherogenic lipoproteins, extracorporeal circuits can also clear non‐lipid plasma constituents. These off‐target effects are clinically relevant in long‐term programs, where cumulative protein losses may influence oncotic pressure, drug disposition, inflammatory tone, and procedural safety [3].

Protein removal during apheresis depends on the technology exploited. In adsorptive systems, like direct adsorption of lipoprotein (DALI), particles bind through ligand and electrostatic interactions and may capture albumin and other proteins [4]. Precipitation‐based systems, such as heparin‐induced extracorporeal LDL precipitation (HELP), alter pH and ionic conditions, with different consequences for co‐precipitation and re‐solubilization [5]. On the other side, membrane‐based approaches, such as double‐filtration plasmapheresis (DFPP), rely on the principle of size‐selective sieving, in which plasma components are separated according to their molecular weight and hydrodynamic radius. In this technique, plasma first passes through a primary membrane filter to remove cellular elements, and is then directed to a secondary filter with defined pore size characteristics that selectively retain larger macromolecules (such as lipoproteins, immunoglobulins, or immune complexes) while allowing smaller solutes such as electrolytes and albumin to pass through and be returned to the patient [6]. Device materials, pore size distribution, flow rates, anticoagulant choice (citrate vs. heparin), and column saturation further modulate non‐lipid protein clearance [7].

Albumin (66 kDa) is the predominant determinant of plasma oncotic pressure, maintaining intravascular volume homeostasis, and serves as the major drug‐binding protein, influencing distribution, free drug levels, and elimination kinetics [8]. Even transient reductions in albumin can perturb fluid balance, tissue perfusion, and pharmacokinetics, with potential downstream effects on hemodynamic stability and therapeutic efficacy [9]. Total protein (TP), though less specific, integrates changes across albumin and the diverse globulin fractions, thereby providing a coarse but clinically useful snapshot of shifts in circulating protein mass and, by extension, plasma oncotic capacity [10]. α2‐macroglobulin (α2M, a 720 kDa tetramer) is a multifunctional acute‐phase glycoprotein that acts as a broad‐spectrum protease inhibitor, a cytokine and growth factor carrier, and a modulator of fibrinolytic and coagulation pathways [11]. Its removal during extracorporeal procedures could theoretically recalibrate inflammatory tone and hemostatic balance, thereby altering both innate immune defense and thrombotic risk [12]. Careful quantification of immediate pre‐ to post‐procedure changes in these markers (and evaluation of how such effects vary according to device characteristics, treatment intensity, and underlying clinical context) is thus directly relevant to patient safety, pharmacotherapy optimization, and long‐term outcomes in routine care.

Until now, the published evidence is heterogeneous regarding the effect of LA on protein levels. Reports differ in timing of post‐apheresis sampling (end‐of‐session vs. delayed draws), reporting units and conversions, concurrent lipid‐lowering regimens, device generations, and clinical indications. Some publications combine multiple methods or filter types within a single cohort, whereas others pool mixed diagnoses, complicating comparison. These design and reporting differences yield inconsistent point estimates and limit the generalizability of single‐center findings.

Herein, we conducted a systematic review and meta‐analysis of observational human studies to estimate pooled, immediate pre‐ to post‐apheresis changes in albumin, TP, and α2M. In addition, the study aimed to identify method‐ and context‐specific modifiers using prespecified subgroups (apheresis duration ≥ 1 year vs. < 1 year, apheresis method, and background disease) and meta‐regression. We also assessed small‐study effects. We hypothesized that LA would produce directionally consistent reductions across all three outcomes; that adsorptive methods would, on average, show larger albumin decreases than precipitation‐ or membrane‐based approaches. Finally, the study attempted to assess whether clinical background would contribute significantly to between‐study variability. Clarifying these patterns can guide monitoring (like timing and frequency of laboratory checks), device and protocol selection for at‐risk patients, and design of prospective studies that link biochemical changes to patient‐centered outcomes.

2. Methods

The present meta‐analysis follows the Preferred Reporting Items for Systematic Reviews and Meta‐analyses (PRISMA) [13].

2.1. Search Strategy

PubMed, Scopus, and Web of Science library were systematically searched from their inception up to August 2025 and studies evaluating the impacts of apheresis on plasma level of albumin, TP, and α2M were included. We composed the search syntax within each database using keywords and common acronyms for (“Lipoprotein apheresis” OR “LDL apheresis” OR “Low density lipoprotein apheresis”) AND (“Albumin” OR “total protein” OR “α2M” OR “α2‐Macroglobulin” OR “alpha‐2‐macroglobulin”). We also manually searched the reference lists of relevant articles to maximize the comprehensiveness of the literature retrieval. To maximize the comprehensiveness of our literature retrieval, no filters were applied during the search process, except for limiting the search to English‐language publications.

2.2. Inclusion and Exclusion Criteria

Publications were eligible to be included if: (i) they were observational studies conducted in humans; (ii) they provided sufficient and convertible data on mean and standard deviation (SD) values or baseline and follow‐up information; (iii) they were published in English; (iv) they at least provided one of three parameters in plasma protein profile (albumin, TP, α2M). Studies were excluded if they were reviews, editorials, letters, case reports, comments, or conference abstracts; or reported data as a percentile of changes instead of mean and SD. When the same population was used in multiple research papers, we selected the most recent study or the one with the largest sample size.

2.3. Study Selection and Data Extraction

Two researchers independently extracted the study data using the same pre‐designed Excel spreadsheet. Any elements that generated controversy were referred to the third researcher for discussion and resolution. The extracted data were: (1) First author name and publication year of the study, (2) study location, (3) the range age or mean age of subjects, (4) participant characteristics, (5) method of apheresis, (6) mean and SD of examined parameters before and after apheresis, (7) sample size, (8) background diseases, (9) the procedures and duration of apheresis, (10) the number of apheresis sessions, (11) medications used concomitantly with apheresis, and (12) timing of post‐apheresis blood sampling. Where reported, the timing of post‐apheresis blood sampling was extracted and categorized as immediate/end‐of‐session sampling or delayed post‐procedure sampling.

2.4. Data Analysis

To evaluate the overall impact of lipid apheresis on plasma protein levels (albumin, TP, α2M), the weighted mean difference (WMD) along with its corresponding 95% confidence interval (CI) was calculated. This meta‐analysis assessed the effect of lipid apheresis on changes in plasma protein levels in participants without a control group. Some studies reported albumin concentrations as medians with interquartile ranges; in these cases, the data were converted to mean ± SD for consistency [14]. The examined factors were reported with a common unit for albumin level (g/L), TP level (g/L), and α2M level (mg/dL). Studies involving a single session of apheresis were excluded from the final analysis. To assess heterogeneity among the included studies, both Cochran's Q test and the I ² statistic were employed. Heterogeneity was considered substantial when the I ² exceeded 50% and the Q test had a p value less than 0.10. Conversely, heterogeneity was considered negligible if the I ² was below 50% and the p value exceeded 0.10 [14, 15]. A random‐effects model (REM) was applied in the presence of significant heterogeneity, while a fixed‐effects model (FEM) was used when heterogeneity was not evident. To explore potential sources of heterogeneity, subgroup analyses and meta‐regression were conducted. Subgroup analyses were based on lipid apheresis duration (≥ 1 year vs. < 1 year), background diseases (hypercholesterolemia, hyperlipoproteinemia, CHD, NS), and type of apheresis method (DALI, HELP, other methods). Meta‐regression analyses were performed to assess the influence of potential confounders, including publication year, apheresis duration, background diseases, and apheresis method. Publication bias was evaluated using Egger's regression test, with a p value below 0.05 indicating significant bias [16, 17]. Sensitivity analyses were conducted by sequentially removing each study to determine its impact on the overall effect size. All statistical analyses were performed using STATA software (version 14.0; StataCorp, College Station, TX, USA) and SPSS (version 23.0; IBM Corp., Armonk, NY, USA).

3. Results

3.1. Literature Search and Study Selection

Figure 1 shows a flowchart of the study selection procedure. A total of 534 potential studies were retrieved; of them, 126 publications were duplicates. After removing the duplicates, 408 studies were retained. According to the inclusion and exclusion criteria, 357 publications by title and abstract and 23 publications by full text examination were excluded. Eventually, 28 publications were eligible for quantitative analysis. Specifically, 22 publications contributed albumin data, 21 contributed TP data, and six contributed α2M data (Figure 1). Among them, some publications utilized different types of filters for apheresis on the same participants; also, there were publications which used different types of apheresis methods that reported data in one paper; some publications reported serum levels of examined factors in different duration times of apheresis. These were treated as separate study‐level comparisons. Publications were conducted between 1991 and 2022, and all of them used lipid lowering drugs accompanied with apheresis therapy. The timing of post‐apheresis protein measurement varied among the studies. Most studies appeared to report values obtained immediately at, or shortly after, completion of the apheresis session; however, delayed or unclearly reported sampling times were also present in several reports. Details on the included studies in terms of publication year, study populations, country of origin, different types of apheresis devices, and other information are presented in Table 1.

FIGURE 1.

Flow diagram of the literature search process for identifying studies.

TABLE 1.

Demographic characteristics of the included studies.

Study author	Publication year	Country	Participants number	Mean ± SD age	Apheresis duration	Background disease	Biochemical enzymes	Plasmapheresis type
Seidel et al. [18]	1991	Germany	46	44.6 ± 8.0	1 year	CHD + hypercholesterolemia	Alb, TP	HELP
Pokrovsky et al. [19]	1994	Russia	3	53	1.5 years	CAD	TP	Immunoadsorption anti‐Lp(a)
Schuff‐Werner et al. [18]	1994	Germany	39	45.5 ± 7.9	2 years	Hyperlipoproteinemia with CHD	Alb, TP	HELP
Suzuki et al. (i) [20]	1996	Japan	5	49.8 ± 14.2	3 months	Heterozygous FH	Alb, TP	DFPP
Suzuki et al. (ii) [20]	1996	Japan	5	49.8 ± 14.2	3 months	Heterozygous FH	Alb, TP	TFPP
Suzuki et al. (iii) [20]	1996	Japan	5	49.8 ± 14.2	3 months	Heterozygous FH	Alb, TP	LDL adsorptive (PA)
Drager et al. [21]	1998	Germany	8	44.8 ± 15.6	5 months	FH	Alb	DALI
Donner et al.* [21]	1996		3				Alb, TP, α2 M	MDF‐LDL apheresis
Hummelsheim et al.*	1995		8				Alb, TP, α2 M	MDF‐LDL apheresis
Muso et al. [22]	1999	Japan	17	NM	3 months	Steroid‐resistant nephrotic syndrome	Alb, TP	Dextran sulfate column (Liposorber LA‐15)
Bosch et al. (i) [23]	2000	Germany	5	64 ± 7	1 months	Hypercholesterolemia with CHD (standard anticoagulant)	Alb, TP	DALI
Bosch et al. (ii) [23]	2000	Germany	5	64 ± 7	1 months	Hypercholesterolemia with CHD (low‐dose citrate anticoagulant)	Alb, TP	DALI
Stenvinkel et al. [24]	2000	Sweden	7	44 ± 7	4 months	Nephrotic syndrome (NS)	Alb	LDL apheresis (Liposorber LA‐15)
Julius et al. (i) [25]	2002	Germany	6	61.5	1 month	FH with CHD	TP, α2 M	HELP
Julius et al. (ii) [25]	2002	Germany	6	61.5	1 month	FH with CHD	TP, α2 M	Lipidfiltration with EC‐50
Suckfüll et al. [26]	2002	Germany	134	49.7 ± 13.5	2 months	SSHL	TP	Fibrinogen/LDL apheresis
Matic et al. (i) [27]	2002	Germany	4	60.25 ± 9.17	2 months	Hypercholesterolemia	Alb	Membrane differential filtration (Asahi filters)
Matic et al. (ii) [27]	2002	Germany	4	60.25 ± 9.17	2 months	Hypercholesterolemia	Alb	Membrane differential filtration (Kuraray filters)
Otto et al. [28]	2003	Germany	10	51 ± 11	3 months	Hypercholesterolemia	Alb, TP	LDL apheresis (Liposorber D)
Blessing et al. (i) [29]	2004	Germany	21	58 ± 3	3–6 years	CHD	Alb, TP	HELP (Secura system)
Blessing et al. (ii)	2004	Germany	21	58 ± 3	3–6 years	CHD	Alb, TP	HELP (Futura system)
Krebs et al. (i) [30]	2004	Germany	20	39 ± 13	15 years (overall cohort)	FH with CHD	Alb, TP	HELP
Krebs et al. (ii) [30]	2004	Germany	20	39 ± 13	15 years (overall cohort)	FH with CHD	Alb, TP	DSA
Krebs et al. (iii) [30]	2004	Germany	20	39 ± 13	15 years (overall cohort)	FH with CHD	Alb, TP	DALI
Krebs et al. (iv) [30]	2004	Germany	20	39 ± 13	15 years (overall cohort)	FH with CHD	Alb, TP	IMAL
Krebs et al. (v) [30]	2004	Germany	20	39 ± 13	15 years (overall cohort)	FH with CHD	Alb, TP	MDF
Klingel et al. [31]	2005	Germany	12	68.6 (57–79)	5 months	Critical Limb Ischemia (CLI)	TP, α2M	DFPP with Rheofilter AR‐2000
Sachais et al. [32]	2005	United States	34	50 (18–67)	2.5 (3 months–7 years)	FH	Alb, TP	LDL apheresis (Liposorber LA‐15)
Tasaki et al. [33]	2006	Japan	31	57 ± 13	2 months	FH and arteriosclerotic diseases	Alb, TP	Liposorber LA‐15
Julius et al. (i) [34]	2007	Germany	6	54.2 ± 7.6	1 month	FH with CAD (Part I)	Alb	DALI 1000
Julius et al. (ii) [34]	2007	Germany	6	54.2 ± 7.6	1 month	FH with CAD (Part I)	Alb	DSA (Liposorber DL‐100)
Julius et al. (iii) [34]	2007	Germany	6	52 ± 11	2 months	FH with CAD (Part II)	Alb	DSA (Liposorber LA‐15)
Julius et al. (iv) [34]	2007	Germany	6	52 ± 11	2 months	FH with CAD (Part II)	Alb	Lipidfiltration with heparin
Julius et al. (v) [34]	2007	Germany	6	52 ± 11	2 months	FH with CAD (Part II)	Alb	Lipidfiltration with citrate
Blaha et al. [35]	2012	Czech Republic	85	66 ± 8	3 months	Dry form of age‐related macular degeneration (AMD)	α2M	Rheohemapheresis
Blaha et al. [36]	2013	Czech Republic	28	66 (54–85)	3 months	Nonvascular age‐related macular degeneration	α2M	Rheohemapheresis
Muso et al. [37]	2014	Japan	44	55.4 ± 17.3 (18–84)	1 month	Drug‐resistant nephrotic syndrome (NS)	Alb, TP	Dextran sulfate column (Liposorber LA‐15)
Gokay et al. (i) [38]	2015	Turkey	10	56.1 ± 2.3	48.6 months	FH	Alb, TP	CF (Cascade Filtration)
Gokay et al. (ii) [38]	2015	Turkey	10	56.1 ± 2.3	48.6 months	FH	Alb, TP	DFPP (Double Filtration Plasmapheresis)
Kozik‐Jaromin et al. (i) [39]	2017	Germany	71	59.6 ± 11.4	21.1 ± 5.7 months	FH, lipoprotein(a)‐hyperlipoproteinemia	Alb, TP	DALI
Kozik‐Jaromin et al. (ii) [39]	2017	Germany	41	57.9 ± 10.1	21.1 ± 5.7 months	FH, lipoprotein(a)‐hyperlipoproteinemia	Alb, TP	MONET
Ishiyama et al. [40]	2021	Japan	32	72.4 ± 6.9	2 years	Cholesterol Crystal Embolism (CCE)	Alb, TP	Dextran sulfate column (Liposorber LA‐15)
Kobayashi et al. [41]	2022	Japan	61	71.5 ± 8 (20–85)	1 year	Chronic limb‐threatening ischemia (CLTI)	Alb	Direct hemoperfusion (AS‐25 device)
Yin et al. (i) [42]	2022	UK and Germany	4	65.75 (60.28–71.22)	3 months	Elevated Lp(a), severe dyslipidemia	Alb, TP	Double‐filtration plasmapheresis+TKM58
Yin et al. (ii) [42]	2022	UK and Germany	8	60.73 (47.5–73.96)	3 months	Elevated Lp(a), severe dyslipidemia	Alb, TP	Double‐filtration plasmapheresis+INUS‐30
Yin et al. (iii) [42]	2022	UK and Germany	5	57.0 (48.07–65.93)	3 months	Elevated Lp(a), severe dyslipidemia	Alb, TP	Double‐filtration plasmapheresis+INUS‐50

Abbreviations: CHD: coronary heart disease, DALI: direct adsorptive lipid‐lowering plasmapheresis, FH: familial hypercholesterolemia.

^*Data of these two studies were reported in the third study and we included the main data but the original article were not available.

3.2. Meta‐Analysis of Albumin

Thirty‐nine studies were included in the present study. In the pooled analysis, the significant decreasing effect of apheresis on serum level (WMD = −3.80 g/L; 95% CI: −4.61 to −2.99; p < 0.001) (Figure 2). Furthermore, subgroup analysis was performed to detect more specific data. Accordingly, the results of analysis based on duration of apheresis therapy demonstrated that both duration < 1 year (WMD = −3.31 g/L; 95% CI: −4.28 to −2.34, p < 0.001, REM) and ≥ 1 year (WMD = −3.72 g/L; 95% CI: −4.93 to −2.52, p < 0.001, REM) of apheresis significantly reduced the level of albumin. We also subanalyzed data based on the methods utilized for apheresis. The results were as follows: DALI (WMD = −5.21 g/L; 95% CI: −6.41 to −4.01, p < 0.001, REM), HELP (WMD = −4.53 g/L; 95% CI: −6.58 to −2.49, p < 0.001, REM), other methods (WMD = −3.34 g/L; 95% CI: −4.25 to −2.43, p < 0.001, REM). We also subanalyzed data based on background disease of participants and interesting data were observed. Accordingly, in most disease subgroups, apheresis decreased serum albumin; the nephrotic syndrome subgroup, however, showed an increase (Table 2).

FIGURE 2.

Forest plot displaying WMD and 95% CI for the effect of apheresis on serum levels of albumin in overall population.

TABLE 2.

The effects of lipid apheresis on plasma proteins in overall population and subgroup analyses.

Parameter		Number of trials	Weighted mean difference	95% CI (p)	Test of heterogeneity
					I 2 (%)	p
Albumin
Overall		39	−3.80	−4.61 to −2.99 (< 0.001)	95.7	< 0.001
Subgroup
Lipid apheresis duration	< 1 years	21	−3.31	−4.28 to −2.34 (< 0.001)	94.7	< 0.001
	≥ 1 years	16	−3.72	−4.93 to −2.52 (< 0.001)	90.3	< 0.001
Background diseases	Hypercholesterolemia	26	−4.31	−5.27 to −3.35 (< 0.001)	96	< 0.001
	Hyperlipidemia (Elevated Lp(a))	4	−5.90	−9.79 to −2.02 (< 0.001)	65.8	0.03
	Nephrotic syndrome	3	4.23	2.62 to 5.84 (< 0.001)	0	0.68
	CHD	2	−6	−7.34 to −4.66 (< 0.001)	0	0.9
Lipid apheresis methods type	HELP	5	−4.53	−6.58 to −2.49 (< 0.001)	92.7	< 0.001
	DALI	6	−5.21	−6.41 to −4.01 (< 0.001)	0	0.93
	Other methods	26	−3.34	−4.25 to −2.43 (< 0.001)	95.3	< 0.001
Total proteins
Overall		34	−8.73	−11.06 to −6.41 (< 0.001)	99.3	< 0.001
Subgroup
Lipid apheresis duration	< 1 years	17	−7.36	−10.67 to −4.05 (< 0.001)	92.4	< 0.001
	≥ 1 years	14	−9.03	−12.58 to −5.49 (< 0.001)	99.7	< 0.001
Background diseases	Hypercholesterolemia	17	−8.53	−10.44 to −6.63 (< 0.001)	91.6	< 0.001
	Hyperlipidemia (Elevated Lp(a))	6	−11.25	−17.99 to −4.5 (< 0.001)	98.6	< 0.001
	CHD	3	−11.29	−12.84 to −9.74 (< 0.001)	0	0.8
Lipid apheresis methods type	HELP	6	−7.86	−13.29 to −2.44 (< 0.001)	98.4	< 0.001
	DALI	4	−11.16	−12.57 to −9.75 (< 0.001)	0	0.9
	Other methods	23	−8.44	−10.67 to −6.22 (< 0.001)	95.5	< 0.001
α2 macro globin
Overall		7	−65.79	−88.98 to −42.60 (< 0.001)	84.3	< 0.001

Abbreviations: CHD: coronary heart disease; CI: confidence interval; DALI: direct adsorptive lipid‐lowering plasmapheresis; ELISA: enzyme‐linked immunosorbent assay; FH: familial hypercholesterolemia.

3.3. Meta‐Analysis of TP

In this analysis 34 studies were eligible to clarify the impact of apheresis on TP. In quantitative analysis, TP levels were significantly decreased in participants compared with pre‐apheresis serum levels (WMD = −8.73 g/L; 95% CI: −11.06 to −6.41; p < 0.001). Sub‐analysis based on duration of apheresis revealed that in both participants who performed < 1 year and ≥ 1 year of apheresis therapy, TP level significantly decreased, respectively (WMD = −7.36 g/L; 95% CI: −10.67 to −4.05, p < 0.001, REM), (WMD = −9.03 g/L; 95% CI: −12.58 to −5.49, p < 0.001, REM) (Figure 3). The results of sub‐analysis according to the background diseases and type of apheresis methods revealed that apheresis reduces serum level of TP (Table 2).

FIGURE 3.

Forest plot displaying WMD and 95% CI for the effect of apheresis on serum levels of TP according to the duration of apheresis.

3.4. Meta‐Analysis of α2M

Overall, seven study‐level comparisons were included in the quantitative analysis of α2M. The pooled analysis showed a significant reduction in α2M following apheresis (WMD = −65.79 mg/dL; 95% CI: −88.98 to −42.60; p < 0.001; REM). However, this estimate should be interpreted cautiously because the evidence base was limited in size and showed substantial heterogeneity (I ² = 84.3%). Subgroup analyses and meta‐regression were not undertaken for α2M because of the small number of available comparisons (Table 2).

3.5. Sensitivity Analysis and Publication Bias

The sensitivity analysis was carried out by excluding the included studies one by one. The results revealed that the included studies had no significant impact on the overall effect size (Figure S3). Furthermore, the funnel plot showed that there was no potential publication bias between included studies except for albumin (Figure S1). The visual findings from the funnel plots were confirmed by quantitative Egger's test for albumin (T = 2.77, p = 0.008), TP (T = 0.89, p = 0.38), and α2M (T = 0.80, p = 0.46).

3.6. Heterogeneity and Meta‐Regression Analysis

Level of heterogeneity of studies was estimated by Cochran's Q‐test and I ² test, and extremely high heterogeneity was observed. Therefore, subgroup analyses were conducted to explore potential sources of heterogeneity. This approach reduced heterogeneity to a nonsignificant level in some subgroups. Furthermore, meta‐regression analysis was also performed to find the potential source of heterogeneity. According to the findings, background p = 0.06 (Figure S2 and Table S1).

4. Discussion

This meta‐analysis demonstrates that LA is consistently accompanied by immediate post‐procedure reductions in albumin, TP, and α2M. The pooled effects were statistically robust across sensitivity analyses, persisted in both shorter and longer treatment programs, and were observed across a range of clinical indications and device types. Between‐study heterogeneity was substantial, yet it decreased within several prespecified subgroups and meta‐regression suggested that background disease may be an important contributor to variability. Evidence of small‐study effects for albumin cautions that some single‐center estimates may be overly optimistic or pessimistic, reinforcing the value of pooled inference.

The clinical interpretation of these magnitudes deserves emphasis. A mean decline of about 3 g/L in albumin following a session will be clinically neutral in most well‐nourished participants, but it can be meaningful in patients whose baseline albumin already approaches thresholds associated with edema, impaired wound healing, or frailty. Because albumin is the dominant determinant of plasma oncotic pressure and a major drug‐binding protein, even a transient decrement after apheresis may shift fluid balance and the free fraction of highly protein‐bound medications [43]. TP decreases amplify this picture by integrating losses across albumin and globulins. The reduction in α2M, a large protease inhibitor and cytokine carrier, raises mechanistic questions about short‐term protease‐antiprotease equilibrium and inflammatory tone [12]. While adverse clinical events were not systematically captured in the included studies, the biochemical pattern argues for vigilance in patients with comorbid malnutrition, chronic inflammation, or procedures scheduled near apheresis sessions.

Differences between apheresis technologies are clinically actionable. Adsorptive systems such as DALI tended to show larger albumin decrements than some alternative approaches, whereas precipitation‐based systems like HELP and mixed or membrane‐selective modalities generally appeared more protein‐sparing. Choice of device and protocol, therefore, should consider not only lipid‐lowering efficacy but also non‐lipid protein preservation, particularly in cohorts at risk of hypoalbuminemia or in programs anticipating long maintenance courses. Practical steps include standardizing anticoagulation strategies, monitoring column saturation and flow parameters that could accentuate non‐specific binding, and revisiting device selection when persistent post‐procedure hypoalbuminemia is observed.

Program duration did not abolish the per‐session effect, as comparable reductions were observed in cohorts treated for less than 1 year and those maintained for a year or longer. This pattern implies that cumulative exposure, rather than adaptation, may dominate the protein trajectory over time. Clinicians should therefore follow trends longitudinally rather than assuming accommodation [44]. The timing of laboratory sampling is crucial; many reports drew blood at the end of a session, whereas others obtained later samples. Without standardized time points, it is difficult to map rebound kinetics or to determine the window during which albumin and α2M are at their nadir. Future studies should incorporate serial post‐apheresis measurements to characterize recovery curves and to link biochemical minima with symptoms and procedural risk.

Clinical background plausibly modifies both baseline levels and the extent of change. Patients with atherosclerotic cardiovascular disease and familial hypercholesterolemia differ from those with nephrotic syndromes or systemic inflammatory states in hepatic synthesis, capillary permeability, and acute‐phase responses, any of which could influence the measured decrements and their recovery. The meta‐regression signal for background disease aligns with this reasoning and suggests that risk‐adapted monitoring (more frequent checks in inflamed or nephrotic phenotypes, for example) may be warranted. Nutritional assessment and targeted support are complementary strategies when persistent low post‐apheresis albumin is observed [45].

Medication management is another implication. For drugs with high protein binding and narrow therapeutic windows, transient albumin reductions may increase the free fraction and, at least theoretically, the pharmacodynamic effect [46]. Although the present evidence cannot quantify these shifts or disentangle them from direct device‐mediated drug removal, it is prudent to review medication lists, consider dosing schedules that avoid administration immediately before sessions when appropriate, and interpret therapeutic drug monitoring with awareness of apheresis timing. Peri‐procedural planning should likewise account for transient protein nadirs; scheduling apheresis away from invasive procedures may reduce the likelihood of edema‐related complications or altered pharmacokinetics of anesthetic agents [47].

In practical terms, clinicians overseeing maintenance LA programs should incorporate structured monitoring of protein profiles at consistent time points, interpret changes in the context of device type and clinical background, and modify protocols when repeated low post‐procedure values emerge. Selecting more protein‐sparing technologies when feasible, optimizing nutrition, and coordinating session timing with surgeries or drug dosing can mitigate off‐target effects without compromising lipid removal [48]. Via aligning device choice, monitoring strategy, and patient‐specific risk, the field can reduce unintended protein loss while preserving the cardiovascular benefits that motivate apheresis in the first place.

This meta‐analysis advances the field by providing pooled estimates in multiple device classes and clinical indications and by showing that immediate post‐procedure reductions in albumin, TP, and α2M are directionally consistent. Nevertheless, several limitations should be considered. First, all included studies were observational and primarily used uncontrolled pre−/post‐apheresis designs; therefore, the observed reductions cannot be attributed exclusively to LA, as peri‐procedural fluid shifts, anticoagulation, hydration status, and concurrent therapies may also have influenced measured protein concentrations. Second, substantial heterogeneity was observed for albumin and TP outcomes (I ² > 90%). To explore potential sources of variability, prespecified subgroup analyses were conducted according to apheresis duration, background disease, and apheresis method. Meta‐regression was additionally undertaken using publication year, treatment duration, background disease, and apheresis method as potential moderators. Although heterogeneity was reduced in selected subgroups, considerable residual heterogeneity remained, indicating that variation in device characteristics, treatment protocols, participant populations, and timing of post‐apheresis sampling likely contributed to the observed differences between studies. Therefore, pooled estimates should be interpreted as average effects across heterogeneous clinical settings. Third, the timing of post‐apheresis blood sampling was not standardized and was incompletely reported in several studies. Most reports appeared to assess protein levels immediately at, or shortly after, completion of apheresis, whereas others used delayed or unclear sampling time points. As protein concentrations may partially recover after treatment because of fluid redistribution and protein re‐equilibration, differences in sampling timing may have contributed to the substantial heterogeneity observed. Accordingly, the present pooled estimates should be interpreted primarily as reflecting early post‐procedure biochemical changes rather than sustained protein depletion. Finally, evidence for α2M was limited, with only seven study‐level comparisons available. This restricted the precision and generalizability of the α2M estimate and precluded robust subgroup or meta‐regression analyses for this outcome. Accordingly, the pooled α2M estimate should be considered preliminary and should not be interpreted as definitive evidence of a consistent effect across apheresis modalities or clinical populations.

5. Conclusion

Based on the available evidence, LA is associated with early reductions in albumin and TP. A decrease in α2M was also observed, but this finding is based on limited and heterogeneous evidence and requires confirmation in larger prospective studies. The directionality is stable, but magnitudes vary by technology and clinical background, suggesting that device selection and risk‐adapted monitoring can mitigate off‐target protein loss without compromising lipid removal. In practice, standardized timing of pre‐ and post‐session laboratory assessment, longitudinal trend review, attention to nutrition, and coordination of apheresis with invasive procedures or dosing of highly protein‐bound medications are reasonable steps while stronger evidence accrues. Future work should prioritize prospective, method‐stratified studies with uniform sampling schedules that map post‐procedure recovery, incorporate therapeutic drug monitoring where relevant, and link biochemical nadirs to clinical outcomes. Such designs will hopefully clarify practical thresholds for intervention and refine technology choices for long‐term programs.

Funding

The authors have nothing to report.

Ethics Statement

The authors have nothing to report.

Consent

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: Funnel plots detailing publication bias in studies included in the meta‐analysis of the effect of apheresis on serum levels of Albumin (A) and total protein (B).

Figure S2: Meta‐regression plots of the effect of apheresis on Albumin. Publication year (A); Apheresis duration category (B).

Figure S3: Leave‐one‐out sensitivity analyses of the impact of apheresis on Albumin (A) and total protein (B).

Table S1: Meta‐regression on the effects of apheresis on potential confounders.

Data Availability Statement

Data sharing not applicable to this article as no datasets were generated or analysed during the current study.