Anti-phosphorylcholine IgM, An Anti-inflammatory Mediator, Predicts Peripheral Vein Graft Failure: A Prospective Observational Study

Abstract

Objectives.

One third of infrainguinal vein bypasses may fail within the first 1¹/₂ years. Pro- and anti-inflammatory mechanisms are thought to be involved in these graft stenoses and occlusions. In previous studies, low levels of anti-phosphorylcholine IgM (anti-PC IgM, an innate anti-inflammatory IgM) have been associated with increased cardiovascular events. In this study we establish the perioperative dynamics of anti-PC IgM levels during leg bypass surgery, and assess the associations between anti-PC IgM levels and primary graft patency.

Design and Methods.

A prospective, observational cohort study of infrainguinal autogenous vein bypass for peripheral arterial occlusive disease. Four university-affiliated hospitals. Plasma cytokine and anti-PC IgM levels were measured pre- and postoperatively. The outcome of interest was loss of primary graft patency due to occlusion or intervention for graft stenosis.

Results.

142 consecutive patients were enrolled: mean age 66 (46–91); 91% white race and male; 72.5% critical limb ischemia (Fontaine III or IV). Median preoperative anti-PC IgM levels were 49 U/ml (IQR 32.3–107.7, mean 89.8 ± 101 std). During follow-up of an average of 1.8 years (1 month to 7.4 years), 50 (35.2%) grafts lost primary patency. Preoperative levels of interleukin-6 or C-reactive protein did not predict graft failure. Patients with preoperative anti-PC IgM values in the lowest quartile had a 2-fold increased risk of graft failure (multivariate Cox proportional hazard, P=0.03, HR=2.11, 95% C.I.: 1.09–4.07), even after accounting for the other significant factors of conduit diameter, distal anastomosis, smoking, and the severity of leg ischemia.

Conclusions.

Low levels of anti-PC IgM are associated with vein bypass graft failure. This biological mediator may be a useful marker to identify patients at higher risk, and offers the potential for novel, directed therapies for vascular inflammation and its consequences.

Brief statement:

In a longitudinal, prospective, observational study of 142 patients undergoing leg bypass with autogenous vein, we found that patients with the lowest preoperative levels of an innate IgM antibody directed against phosphorylcholine (anti-PC) were twice as likely to suffer loss of primary patency, after controlling for other risk factors of graft failure. This naturally occurring biological mediator may be a useful marker to identify patients at higher risk of graft failure, and offers the potential for novel, directed therapies for vascular inflammation and its consequences.

INTRODUCTION

Pathologic inflammation within the arterial wall has long been implicated in chronic peripheral arterial disease (PAD), as well as in the failure of endovascular interventions and vein grafts¹. Although statins are known to have pleiotropic anti-inflammatory effects², much less is known about innate anti-inflammatory mechanisms that can attenuate these pathologic responses to vessel wall injury. The human innate immune system includes a family of antibodies that naturally recognize danger-associated molecular patterns (DAMPs), which are localized inflammatory signals liberated by oxidized phospholipids, and by cell injury and apoptosis^{3, 4}. Among these DAMPs is the polar headgroup (phosphorylcholine or PC) of the membrane phospholipid phosphatidylcholine. PC is a common inflammatory component of damaged cell membranes⁵. This normally hidden phospholipid is exposed during cell injury and death, and serves to instigate a cascade of local, inflammatory cellular events⁶. As a component of the innate immune system, humans have a naturally occurring IgM antibody that specifically targets PC⁷. This antibody is believed to play an anti-inflammatory role, by attenuating the inflammatory response to the danger signal, and facilitating clean up and repair at sites of cell injury. Low levels of this antibody have been associated with an increased risk of myocardial infarction and stroke in large population-based studies^8–10. Likewise, studies suggest that higher levels of this antibody may attenuate the inflammatory response to injury: these antibodies reduce macrophage activation in vitro^{8, 11}, and reduce intimal hyperplasia in animal models¹².

To date, these innate, anti-inflammatory antibodies have been mainly studied in the context of chronic atherosclerosis. Little is known about their role in the vascular response to injury, for example in peripheral arterial vein grafts. In a previous pilot study¹³ we found a positive relationship between low levels of this anti-PC IgM and peripheral vein graft failure. To expand on this initial finding, we have now completed a larger, longitudinal, prospective, observational cohort study of patients undergoing lower extremity vein graft bypass for arterial occlusive disease. The goals of this study were to establish the dynamics of anti-PC IgM levels throughout the perioperative period, and to determine if the plasma levels of this anti-inflammatory IgM antibody correlate with long term graft patency – freedom from stenosis and occlusion.

MATERIALS AND METHODS

A Prospective Observational Cohort Study.

This was a prospective, observational, longitudinal cohort study of patients undergoing infrainguinal bypass with autogenous vein. All subjects were recruited from the VA Puget Sound Health Care System and the University of Washington hospitals. Recruitment was between 2008 and 2016, and the study end date was 31 December, 2017. The procedures and protocols were approved by the Institutional Review Boards for human studies at the University of Washington, and the VA Puget Sound Health Care System, Seattle, WA. All subjects gave their informed consent.

Eligibility.

All consecutive subjects scheduled for elective infrainguinal bypass with autogenous vein for symptomatic peripheral arterial occlusive disease were eligible, were identified from surgical schedules, and recruited before surgery.

Exclusions.

Subjects were not recruited if they were on hemodialysis, had a confirmed diagnosis of any autoimmune disease (e.g. rheumatoid arthritis or lupus), or any diagnosed prothrombotic diathesis, or if the operation was for peripheral aneurysmal disease. Subjects were excluded postoperatively if the planned autogenous bypass was not performed (i.e. prosthetic graft), if the bypass was combined with open surgical bypass of the iliac arteries or aorta, or if no satisfactory intraoperative completion imaging was performed. Grafts that occluded within the first 30 days after surgery were also excluded, because these failures are primarily technical in nature¹⁴.

Follow-up.

Patients were followed regularly at clinically prescribed intervals to assess for graft patency. Standardized duplex ultrasound surveillance examinations of the graft were performed at 1, 3, 6, 12, 18 months, then yearly thereafter, according to the procedures and principles outlined by Bandyk¹⁵. Their published guidelines for critical graft stenoses were followed (peak systolic velocity >300 cm/s, velocity ratio > 3.5), but the final decision for graft intervention was left to the judgement of the treating surgeon. The treating clinicians were blinded to the results of our IgM and cytokine measurements. Follow-up indices were calculated for all subjects according to von Allmen et. al.¹⁶. The IgM quartile was not associated with cohort attrition by regression analysis, so losses to follow-up were censored in Cox proportional hazards analysis.

Endpoints, Covariables.

The outcome of interest was loss of primary patency. This was defined as “graft failure” (endpoints were either thrombosis, or a surgical or endovascular intervention for a critical vein graft stenosis). After primary graft failure, follow-up was ended, and secondary or assisted patency was not tallied.

The clinical indications (i.e. original rationale) for surgery were classified according to the guidelines of the European Society for Vascular Surgery¹⁷: claudication (Fontaine stage II, Rutherford Grade I), ischemic rest pain (Fontaine stage III, Rutherford Grade II) or critical limb ischemia with tissue loss (Fontaine stage IV, Rutherford Grade III). The severity of leg ischemia was categorized prospectively using the WIfI scale^{17, 18}, measuring all three modalities (ankle-brachial index, absolute ankle pressure, and toe pressure/TcPO₂) in each patient . For simplicity, throughout this report the WIfI score for ischemia is referenced according to the ankle-brachial index (ABI): Grade 1 – ABI = .6–.79, Grade 2 – ABI = .4–.59, Grade 3 – ABI < .4, even though all three WIfI parameters for ischemia were used to categorize their ischemia. No subjects were in Grade 0.

The clinical covariables listed in Table 1 were extracted from the electronic medical records prior to surgery. Antiplatelet therapy was defined as the current use of any one or combination of aspirin, thieneopyridine drug, or dipyridamole. Operative variables included the length of the venous conduit, the location of the distal anastomosis, and the minimum diameter of the distended vein before implantation. These data were derived from a contemporaneous survey completed by the surgeon. The methods for harvesting the donor vein were left to the discretion of the surgeon (see Results), although all surgeons utilized the recognized standards of practice, avoiding excessive dissection of the vein and overdistention. Endoscopic harvesting techniques were not used.

Table 1:

Patient and Surgery Characteristics

Characteristic	Number	Percent or Interquartile Range
Total Subjects	142
Follow-Up (median days)	480	171–903
Follow-Up Index	0.68	0.5–1.0
Age (median years)	65.5	60–70
White Race	129	90.8%
Male Sex	129	90.8%
Current Smoker	71	50%
Diabetes	67	47.2%
Conduit using Greater Saphenous	136	95.8%
Length of conduit (median cm)	46	37–55
Min vein diam (median mm)	4	3.5–4.5
Distal Anastomosis
Popliteal	56	39.4%
Tibial	86	60.6%
Critical Limb Ischemia (Fontaine III or IV)	103	72.5%
WIfI Ischemia Score 2 or 3 (AAI < .6)	105	73.9%
Statin therapy	116	82.3%
Anti-platelet therapy	129	90.8%
Preop Anti-PC IgM (median units/mL)	48.98	32.2–107.9
Preop IL-6 (median pg/mL)	7.2	1.6–20.9
Preop C-reactive protein (median mg/L)	17.2	5.2–55.4

Statistical Design.

Previous clinical studies have indicated that the lowest levels of anti-PC IgM are closely linked to adverse cardiovascular outcomes, while a much broader range of average or high levels is associated with freedom from risk^8–10. Thus, for this study we chose, in advance, to compare the cohort of patients in the lowest quartile of anti-PC IgM to those in the highest three quartiles. Based on the median graft survival times from our pilot studies¹³, we calculated that it would require a total sample size of 108 patients to reject the null hypothesis that the graft survivals of the lowest quartile and highest three quartiles are equal, with a probability (power) of .8 and a Type I error of 0.05.

Statistical analyses were performed using the IBM SPSS software (version 24, Armonk, NY). Group categories were compared by the Pearson chi-square and t-tests. Hazard ratios for loss of primary patency were estimated by univariate and multivariable Cox proportional hazards tests and survival curves. Plasma cytokine values were log-transformed for analysis, when they were not normally distributed. The defining level for the lowest quartile of IL-6 was 1.6 pg/mL, and for CRP 5 mg/L. Missing values for cytokine and IgM levels were handled by pairwise deletion.

Plasma Measurements.

Blood samples were obtained preoperatively, on postoperative day 1, day 3–5, then at postoperative months 1, 3, and 6. Standard assays were employed (see Supplementary Materials).

RESULTS

Characteristics of Study Population.

One hundred and eighty-six consecutive subjects were recruited preoperatively, and 44 were excluded by the postoperative exclusion criteria (e.g., prosthetic graft). One hundred and forty-two eligible subjects were followed for an average of 1.8 years (1 month to 7.4 years). Table 1 summarizes the clinical characteristics of this population, their operations, and follow-up indices. Sixty-one percent of the bypasses were made to the tibial arteries (39% popliteal), and 96% used the greater saphenous vein (4% arm veins). The surgical harvesting technique was resection and re-implantation in all but three cases, where the in situ technique was used. During their postoperative hospitalization, 72% of patients received subcutaneous, prophylactic doses of heparin. Regarding cytokine and IgM values, no preoperative values were missing, and 5–7% of postoperative values were missing. These were missing completely at random (MCAR) in relation to IgM level or outcome, and were handled by pairwise deletion in statistical analyses.

Associations Between IgM Levels and Clinical Characteristics.

Table 2 compares the main clinical characteristics and follow-up parameters for the lowest quartile of preoperative anti-PC IgM versus those in the higher three quartiles. There was a higher proportion of tibial bypasses in the high 3 quartile cohort, but otherwise there were no significant differences in the distribution of characteristics between the two groups.

Table 2.

Characteristics of Subjects in the Lowest versus Highest 3 Quartiles of Anti-PC IgM

	Lowest Quartile		Highest 3 Quartiles
Characteristic	Number or Value	Percent or IQR	Number or Value	Percent or IQR	P value*
Preoperative Anti-PC IgM (median u/mL)	23.5	18.1–27.7	61.1	40.7–137.9
Follow-up Index (median)	0.826	0.37–1.0	0.663	.51–1.0	0.88
Preoperative IL-6 (median pg/mL)	14.9	1.8–30.1	6.6	1.5–20.2	0.1
Preoperative CRP (median mg/L )	21.4	9.2–47	15.7	4–56.4	0.53
Age (median)	68	62–75	65.3	60–69	0.12
White Race	29	87.9%	100	91.7%	0.31
Male Sex	30	90.9%	99	90.8%	0.99
Current Smoker	13	39.4%	58	53.2%	0.16
Diabetes	18	54.5%	49	45.0%	0.33
Length of conduit (median cm)	47.6	38–60	45.5	37–51	0.73
Min vein diam (mean mm)	3.9	3.5–4	4	3.5–4.5	0.56
Distal Anastomosis
Popliteal	19	57.6%	37	33.9%	0.026
Tibial	14	42.4%	72	66.1%
Critical Limb Ischemia (Fontaine III or IV)	27	81.8%	76	69.7%	0.17
WIfI Ischemia Score 2 or 3 (AAI < .6)	25	75.8%	80	73.4%	0.79
Statin therapy	28	84.8%	88	81.5%	0.66
Anti-Platelet therapy	30	90.9%	99	90.8%	0.99

IQR=Interquartile range

^*P values determined by Pearson chi-square or t-test

Because the preoperative clinical conditions of the patient (e.g. diabetes, Fontaine stage, smoking) might influence the baseline IgM level, these associations were examined by univariable linear regressions, and then by multiple stepwise regression (supplemental Table 1). This revealed that only a patient’s clinical indication for surgery (i.e., claudication vs. critical limb ischemia, Fontaine II vs. III-IV) bore a significant association with the preoperative IgM level, albeit with weak effect (R²=.04, F(1,141)=7.18, P<0.01). More severe clinical presentations tended to have lower IgM levels. Thus, we included this variable in our subsequent analyses of the relationship between IgM level and graft survival.

Clinical and Biologic Factors Associated with Primary Graft Failure.

Primary endpoints occurred in 50 subjects (35.2%, including 12 graft thromboses and 38 interventions for stenosis). Their median time to failure was 181 days. Because many confounding factors may influence graft patency, we individually examined each of the the potential clinical and biochemical variables that might be associated with graft patency, using univariate Cox proportional hazards analysis.

Figure 1 shows the leading factors that were significantly associated with primary graft failure in a univariable fashion (P≤0.05): the location of the distal anastomosis, the diameter of the arterialized conduit, the lowest quartile of preoperative anti-PC IgM levels, and the degree of ischemia as classified by the WIfI scale. Borderline associations with graft survival (P between 0.05 and 0.2) were found for: Fontaine Class, conduit length, smoking status, and a preoperative CRP > 5 mg/L. Age, diabetes, and preoperative IL-6 levels > 1.6 pg/mL were not associated with graft failure.

Figure 1. Clinical and Biochemical Factors Associated with Loss of Primary Patency.

Univariable Cox regression hazard ratios of clinical factors affecting primary graft patency. Errors bars indicate 5%−95% confidence intervals. Distal anastomosis, vein diameter, the lowest quartile of anti-PC IgM, and a WIfI ischemia score of 2 or 3 were associated with a significantly higher risk of graft failure (P≤0.05). Of borderline significance (P<0.2) were Fontaine class, conduit length, current smoking, and CRP>5 mg/L.

The eight variables found to be individually associated with graft survival at a level of P = 0.2 or less were then included in a Cox multivariable model. Backwards stepwise regressions yielded the key significant factors associated with graft survival. Table 3 shows the final analysis: a patient with an anti-PC IgM level in the lowest quartile experienced a 2-fold greater risk of graft failure, even after accounting for the distal anastomosis, conduit diameter, smoking, and the WIfI ischemia category. Figure 2 compares the Cox model of graft survivals of the lowest and high 3 quartiles of anti-PC IgM, after accounting for the other confounding variables in Table 3.

Table 3.

Clinical and Biochemical Factors Associated with Loss of Primary Graft Patency.

Clinical Factor	Hazard Ratio*	95% Confidence Interval		P value
		Lower	Upper
Tibial Distal Anast. (vs. Popliteal)	2.40	1.35	4.27	0.003
Vein Diameter (below median)	2.53	1.18	5.43	0.02
Lowest Qrtile IgM	2.11	1.09	4.07	0.03
Current Smoker	1.82	1.01	3.29	0.05
WIfI Ischemia Score 2 or 3 (ABI<.6)	2.03	0.96	4.27	0.06

^*Multivariable Cox Proportional Hazards

Figure 2. Cox Survival Plot of Primary Graft Patency.

Graft survival of the subjects with the lowest quartile of preoperative anti-PC IgM levels versus those in the highest three quartiles, after accounting for the significant clinical factors in Table 3. The time axis is truncated at the point at which standard errors exceed 10%.

Perioperative Dynamics of IgM Levels and the Inflammatory Response

IgM levels fell in concert with the acute rise in systemic inflammation provoked by the surgery. Figure 3 illustrates these changes, and compares the change in IgM levels with the corresponding changes in two classical measures of acute inflammation, interleukin 6 (IL-6) and C-reactive protein (CRP). We were also interested to see if the patients with the lowest levels of anti-PC IgM might exhibit different patterns of postoperative inflammatory response. The postoperative fluctuations in IL-6 and CRP levels were no different between the cohorts with the lowest quartile of preoperative IgM levels versus the highest three quartiles. As previously noted in Table 3, there was no significant association between the preoperative levels of anti-PC IgM and those of IL-6 or CRP.

Figure 3. Time course of dynamic changes in Anti-PC IgM, Interleukin-6 (IL-6), and C-reactive protein (CRP).

* P < 0.05, t-test comparing postoperative values to the preoperative baseline. Abbreviations: POD = postoperative day, mo = month.

DISCUSSION

This longitudinal, prospective study of 142 patients shows that low plasma levels of a naturally occurring anti-inflammatory factor, anti-phosphorylcholine IgM, bear a strong relationship to bypass graft failure, even after accounting for other clinical, biochemical, and technical factors. Those patients in the lowest quartile of preoperative IgM levels had a 2-fold higher risk of primary graft failure, controlling for the other significant factors of distal anastomosis, conduit diameter, smoking, and the severity of leg ischemia. The clinical characteristics of the two groups were generally similar, although the high 3 quartile group had a smaller proportion of popliteal bypasses. Because femoral-popliteal bypasses generally have a superior patency to tibials, this inequality in distribution would favor an improved graft survival in the low IgM group, while we found the opposite. The effects of anti-PC IgM levels observed here support prior studies that suggest that a wide range of normal or high levels of this anti-inflammatory antibody may be protective against cardiovascular complications, while low levels are associated with higher risk. Supporting this, we did not find significant differences in graft survival between those subjects in the second, third, or highest quartiles (not shown).

In the population studied here, the rate of primary graft failure (35.2%) was within the range reported in the literature (25–35%)¹⁹. Also, the other clinical characteristics that we found to be associated with stenosis and graft failure (graft diameter, distal anastomosis, smoking, ischemia category, Fontaine/Rutherford class) have been well documented as risk factors in previous studies of graft patency^20–24.

Patients with peripheral arterial disease are well known to have significantly higher levels of chronic systemic inflammation, compared to those without arterial disease^25–27. And an unregulated inflammatory response to injury has long been implicated as a culprit in the process of intimal hyperplasia and its range of manifestations: from focal, stenotic lesions to a more generalized failure of adaptive remodeling^{19, 28}. The patients in this study experienced a typical inflammatory response after surgery, as reflected by increases in IL-6 and CRP, while anti-PC IgM levels fell (Figure 3). But the postoperative inflammatory response did not appear to be influenced by their initial levels of anti-PC IgM (Figure 4). Nor were the high preoperative plasma levels of IL-6 and CRP predictive of subsequent graft failure (see Figure 3).

Figure 4. Comparison of the dynamic changes in Interleukin-6 (IL-6), and C-reactive protein (CRP) according to the preoperative value of Anti-PC IgM.

Abbreviations as in figure 3.

Indeed, few studies have succeeded in identifying any baseline inflammatory marker that might accurately predict the patency in peripheral arterial bypass. Owens and colleagues observed that higher baseline levels of CRP were associated with impaired luminal remodeling²⁹ in humans, but did not correlate this with outcomes. They also observed a higher rate of graft failure in women with baseline CRP > 5 mg/L, but not in men³⁰. We also found no association between baseline CRP or IL-6 and later graft failure in our predominantly male population. Stone and colleagues found an association between two markers (CRP and brain natriuretic protein), and overall adverse outcomes after peripheral endovascular interventions, but no relationship to patency³¹. In a recent study of gene activation and proteomic changes after peripheral endovascular procedures, DeSart et. al. further expanded the list of potentially predictive markers, implicating more complex patterns of proteomic and genetic changes associated with the failure of vascular interventions³². Taken together, these data suggest that the pro-inflammatory milieu in the postoperative period is more complex than can be accounted for by one or two markers.

Because the levels of this innate anti-inflammatory antibody did not appear to modulate the acute, systemic inflammatory response to vascular surgery (as least as reflected by these markers), one might speculate that the anti-inflammatory actions of anti-PC IgM could be due to local actions within the vessel wall where this danger-associated molecular pattern (phosphorylcholine) is exposed by injured and apoptotic cells. The consistent drop in plasma anti-PC IgM levels seen early postop (Figure 3) may be due, at least in part, to the consumption of IgM by tissues exposing phosphorylcholine.

This current study has several limitations. The study population consisted predominantly of Caucasian males, so we cannot draw any conclusions about women, or other races. Our findings are not applicable to endovascular interventions or prosthetic grafts, which were not studied. Because this this was an observational study, surgical techniques and postoperative antithrombotic therapy were not standardized. However, the surgical characteristics of the procedures and their postoperative care were relatively uniform: 96% greater saphenous, 98% resected and re-implanted. Eighty to 90% of the patients in the study were under treatment with a statin and an antiplatelet agent, and three-quarters received postoperative prophylactic doses of heparin. Confident discrimination of the effects of these factors could not be made because of the small numbers of patients who were not receiving aspirin or a statin, for example.

Finally, there is an inherent bias in our late (3–6 month) measurements of IgM and the cytokines. Our protocols did not measure cytokines and IgM levels after subjects suffered primary graft failure (median time to failure ~ 6 months), because we felt the measurements would be confounded by the events of thrombosis and/or re-intervention. Thus, the late time points of three and six months in Figure 3 reflect the attrition of those with failed grafts. However, the earlier timepoints in the acute perioperative period are more complete, and representative of the entire cohort.

This current study emphasizes the potential value of innate anti-inflammatory mechanisms such as the anti-PC IgM antibody, which acts through a variety of natural anti-inflammatory mechanisms, including the clearance of oxidized LDL and damaged cells, and recruitment of classical complement pathways^{6, 7, 11, 33}. Current guidelines of the European Society for Vascular Surgery¹⁷ recommend single agent antiplatelet therapy after surgical revascularization, and this current study should not change that. However, if future studies studies confirm this association between low levels of anti-PC IgM and graft failure, then preoperative diagnostic testing could identify those who might benefit from closer surveillance and/or more intensive antithrombotic therapy. In the future, therapy via passive or adaptive immunity may also be possible.