Diagnostic Performance and Prognostic Value of Extravascular Retention of ¹²³I-Labeled Serum Amyloid P Component in Systemic Amyloidosis

Patients

All 216 consecutive new patients with amyloidosis who were evaluated at the University Medical Center Groningen, a tertiary referral center, from February 1990 until December 2003 were prospectively screened for the study and subsequently followed until July 2006. Amyloidosis was diagnosed in all patients by the presence of typical apple-green birefringence in polarized light in a tissue specimen stained with Congo red dye. The definition and characterization of localized and systemic amyloidosis, the classification of systemic amyloidosis into AA, AL, and ATTR types, and the clinical evaluation of heart, liver, and kidney involvement have been described in detail elsewhere (10). Peripheral nerve involvement was defined to be present clinically if so judged by a consultant neurologist. Autonomic nerve involvement was defined to be present when the modified Ewing test battery yielded abnormal results (15). Twenty-five patients were not included: 10 because of logistical problems, 9 because of sudden death, and 6 because they did not want to participate. Twenty controls were studied. They were patients who had diseases that can underlie amyloidosis but in whom biopsies for amyloid had been negative and no features suggesting amyloidosis had developed during a follow-up of 3–9 y. The local Ethics Committee approved the study, and all 189 patients and 20 controls included gave informed consent to participate. The population used in this study was identical to an earlier-described population in which the diagnostic performance of ¹²³I-SAP scintigraphy was studied (10).

Radiolabeling and Quality Control

From 1990 to 1999, highly purified human SAP was obtained from London (kindly provided by Professor Philip Hawkins) (4). Since 1999, SAP has been purified and prepared from Dutch blood donors according to Dutch pharmaceutical standards in our center in Groningen. An essential part of these standards was the extensive screening of the donor blood for infectious diseases, including HIV. In all donors, this screening was repeated 6 mo after blood donation. Radiolabeling with radioactive iodine (¹²³I, half-life of 13.2 h, obtained from GE Healthcare [formerly Amersham Cygne]) and quality control were performed as described elsewhere (12).

Dose Administration, Blood Sampling, and Visual Assessment of Scintigraphy

Tracer administration and blood sampling were performed as reported earlier, with few modifications (12). In short, 200 MBq of ¹²³I-SAP containing 100 μg of protein were given as an intravenous bolus. The dose used was chosen for imaging. Venous plasma samples were drawn just before injection and 10 min, 4 h, 24 h, and 48 h after injection. Urine was collected for the first and second 24 h after administration. Thyroid uptake of free iodide was prevented by oral administration of potassium iodide. Subjects with a history of adverse reactions to iodide or to intravenous radiologic contrast media were excluded before administration.

Scintigraphy was performed 24 h after injection and assessed visually as described elsewhere (10). In short, total-body anterior and posterior views, as well as abdominal anterior and posterior spot views, were acquired. Two investigators who were blinded to the clinical data assessed normal or increased organ uptake (graded from 1+ to 3+) in patients with amyloidosis and controls. The highest organ score present in the patient determined the body uptake score.

SAP Retention Studies

One milliliter of each plasma sample and 2 mL of the urine portions were counted. Trichloroacetic acid–precipitable (protein-bound) radioactivity in plasma was used to measure the percentage of the injected dose still present in the vascular compartment, by using a calculated estimate of the plasma volume (16). The blood volume (in mL) was calculated by applying an algorithm using sex, body weight (in kg), and height (in m) (17): male blood volume = (366.9 × height³) + (32.19 × weight) + 604, and female blood volume = (356.1 × height³) + (33.08 × weight) + 183.3. Whole-body hematocrit was calculated from the venous hematocrit by multiplying by 0.91 (18). Plasma volume was calculated by subtracting the calculated red cell mass from the blood volume: plasma volume = blood volume – 0.91 × venous hematocrit × blood volume. The calculated plasma volume measurements were compared (Bland–Altman analysis) with the extrapolated distribution volumes at the moment of injection, as had been assessed previously (12) in 75 patients: After correction for the influence of 4 extreme outliers, the mean bias between distribution volume at injection and plasma volume was 11.7% (SD, 25.9%). Compared with other possible approximations, this calculation of the plasma volume appeared to be the closest to the extrapolated distribution volume. The plasma volume—multiplied by 1.12 to correct for the bias—was used to estimate the distribution volume of SAP, resulting in the following equation: distribution volume of SAP = 1.12 × plasma volume = 1.12(blood volume – 0.91 × venous hematocrit × blood volume). For practical use and simplicity, the ratio of the injected dose to the distribution volume of SAP was the concentration of radioactivity at the moment of injection and was set to 100%. Plasma samples (at 10 min and 4, 24, and 48 h) were thus expressed as percentage of injected dose. ¹²³I-SAP disappearance from the circulation was also measured independently of the distribution volume by calculating the ratio of plasma retention after 4 h to that after 10 min. This ratio was called the plasma SAP disappearance index (PSDI).

Because there is no significant extrarenal excretion, the total-body retention was calculated by subtracting the cumulative urine radioactivity, expressed as a percentage of the injected dose, from the given dose. Total-body retention values after 24 and 48 h were thus calculated. Whole-body counting during scintigraphy may be an alternative way to measure total-body retention (12). Extravascular retention values after 24 and 48 h (EVR₂₄ and EVR₄₈) were determined by subtracting the total-plasma radioactivity (expressed as a percentage of the injected dose) from the total-body retention (12).

Statistical Analysis

Statistical analysis was performed using the statistical package GraphPad Prism, version 4.02 (GraphPad Software Inc.). The Bland–Altman analysis was used to compare the distribution volume at injection with the calculated plasma volume. The 1-way ANOVA test was used in combination with the Bonferroni multiple-comparison test to detect differences in variables among groups. Receiver-operating characteristic curves were used to analyze the different possible diagnostic cutoff points for SAP retention studies. The Pearson test was used to detect correlations. Survival curves were constructed according to the Kaplan–Meier method, and the groups were compared using log-rank tests. In all tests, 2-tailed P values of less than 0.05 were considered significant.

SPSS 12.0.1 (SPSS Inc.) for Windows (Microsoft) was used for predictors of survival. We identified predictors of survival in univariate and multivariate Cox proportional-hazard models and calculated the relative hazards and 95% confidence intervals. Multivariate analyses were performed with a stepwise forward-regression model, with an entry probability for each variable set at 0.05.

Patient Characteristics

One hundred eighty-nine patients were included in the study: 60 with AA, 80 with AL, and 27 with ATTR amyloidosis and 22 with localized disease (10). In addition, 19 patients without amyloid (8 with chronic arthritis, 3 with plasma cell dyscrasia, 6 with transthyretin mutations, 1 with familial Mediterranean fever, and 1 with polyneuropathy) and 1 healthy person served as controls. Patient characteristics and organ involvement for the 4 most prominently affected organs are shown in Table 1. Patients with AL amyloidosis were older (P < 0.05) than controls. As expected, cardiac involvement (55%) and liver involvement (30%) were most prominent in AL amyloidosis. Kidney involvement was most prominent in both AA (90%) and AL types (83%) and was also frequently part of the clinical picture of the controls without amyloidosis (55%). Nerve involvement was most prominent in ATTR (85%) but was also frequent in AL amyloidosis (50%). Patients with the AL type most frequently had involvement of more than 2 of the 4 organ systems (39%). Visceral organ disease was judged to be a coincidental finding in 2 patients with localized amyloidosis (nerve and kidney), and in both cases amyloid deposition in the affected tissue and organ was excluded.

Diagnostic Sensitivity and Specificity

Table 2 shows retention values of plasma, total body, and extravascular ¹²³I-SAP at the 4 intervals after injection. Patients with AA and AL amyloidosis had lower plasma retention 4 h after injection, as well as higher total-body and extravascular retention 24 and 48 h after injection, compared with controls. Patients with ATTR and localized amyloidosis did not differ from controls.

Receiver-operating characteristic curves were used to assess diagnostic cutoff points for the ¹²³I-SAP retention studies. All patients with systemic amyloidosis were merged into a “systemic amyloidosis” group, and controls and patients with localized amyloidosis were merged into a “nonsystemic amyloidosis” control group to calculate diagnostic specificity for systemic amyloidosis. Analysis of plasma retention measurements showed that PSDI discriminated best, with an area under the curve of 0.73. When a cutoff value of 0.51 was chosen for PSDI, reflecting a desired specificity of 90%, the sensitivity figures turned out to be 30% for AA, 38% for AL, and 11% for ATTR amyloidosis. Analysis of body retention measurements showed that EVR₂₄ discriminated best, with an area under the curve of 0.83. When a cutoff value of 50% was chosen for EVR₂₄, reflecting a desired specificity of 90%, the sensitivity figures turned out to be 65% for AA, 61% for AL, and 22% for ATTR amyloidosis. The combination of the 2 parameters using their cutoff values (PSDI > 0.51 and EVR₂₄ < 50%) produced an associated specificity of 86% and associated sensitivities of 70% for AA, 63% for AL, and 26% for ATTR amyloidosis.

All 167 patients with systemic AA, AL, or ATTR amyloidosis were subdivided on the basis of the intensity of scintigraphic uptake from 0 to 3+. Figure 1 illustrates the sensitivity figures of 18%, 44%, 53%, and 92% for these patients with scores of 0, 1+, 2+, and 3+, respectively (i.e., the percentage of patients in each group with values outside the combined reference area).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1. 

PSDI and EVR₂₄ of all 167 patients with systemic amyloidosis (AA, AL, and ATTR types). Patients are subdivided according to intensity of scintigraphic uptake, represented by × (negative, n = 28), ○ (score 1+, n = 45), (score 2+, n = 32), and • (score 3+, n = 62). Dotted lines are reference limits of controls (>0.51 for PSDI and <50% for EVR₂₄), resulting in reference area in upper left quadrant.

EVR₂₄ and Organ Involvement

The EVR₂₄ was higher in AL (P < 0.001) and ATTR (P < 0.05) patients with heart involvement (means of 64% and 45%, respectively) than in such patients without heart involvement (means of 47% and 30%, respectively). The EVR₂₄ did not differ between patients with and without nerve involvement in any of the amyloid groups.

In AL amyloidosis, the EVR₂₄ correlated modestly well with serum alkaline phosphatase (r = 0.63, P < 0.0001, Fig. 2A) after log transformation and correlated weakly with the endogenous creatinine clearance (r = −0.36, P < 0.005, Fig. 2B). In AA amyloidosis, the correlation of EVR₂₄ with these liver and kidney function tests was poor (r = 0.11 and r = −0.21, respectively) and statistically not significant. In all groups with amyloidosis, no correlation was found between EVR₂₄ and proteinuria.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2. 

Relationship between EVR₂₄, serum alkaline phosphatase (A), and endogenous creatinine clearance (B) in patients with AL amyloidosis. Solid lines are linear regression lines.

The mean EVR₂₄ increased significantly (P < 0.001) with the number of organs involved in AL amyloidosis (from 43% for 1 organ to 81% for 4 organs), whereas this increase was not found in AA amyloidosis. The number of ATTR patients was too small to show differences, but a linear trend was present (P < 0.05).

Survival Analysis

After the ¹²³I-SAP retention study, the median follow-up of the 167 patients with systemic amyloidosis was 54 mo. Lost to follow-up were 5 AL patients (after a median of 3 mo), 12 AA patients (after a median of 20 mo), and 1 ATTR patient (after 11 mo). Fifty-five AL patients, 38 AA patients, and 12 ATTR patients had died. Median survival differed among the groups (P < 0.005): 12 mo in AL patients, 37 mo in AA patients, and 134 mo in ATTR patients.

Univariate survival analysis was performed on all groups of patients during the first 5 y after the study. Significant results were seen only in patients with AL amyloidosis, as shown in Table 3. A significantly elevated hazard ratio was found for elevated EVR₂₄, underlying multiple myeloma, heart involvement, and number of organs. Multivariate survival analysis in AL amyloidosis showed that elevated EVR₂₄ still remained a predictor of survival (hazard ratio, 2.0) after clinical heart involvement (hazard ratio, 3.9) or heart failure (hazard ratio, 5.0) had first been entered, as shown in Table 3, whereas the number of organs and underlying multiple myeloma lost their significance as predictors of survival.

The predictive role of both cardiac involvement and EVR₂₄ is shown in Figure 3. Figure 3A demonstrates the survival curves of AL amyloidosis patients with cardiac involvement (median survival, 6 mo) and without cardiac involvement (median survival not reached after 60 mo). Figure 3B demonstrates the survival curves of AL amyloidosis patients with elevated EVR₂₄ (median survival, 6 mo) and normal EVR₂₄ (median survival, 59 mo). Figure 3C demonstrates the survival curves for the group with or without cardiac involvement further subdivided into patients with normal or elevated EVR₂₄: The effect of EVR₂₄ on survival was particularly evident in the group with cardiac involvement (median survival was 4 mo for members of this group with elevated EVR₂₄ and 25 mo for members of this group with normal EVR₂₄; P < 0.05), whereas this effect of EVR₂₄ was not found in the group with normal cardiac function (median survival had not been reached for either group by 5 y).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3. 

Five-year survival of 80 patients with AL amyloidosis. (A) Patients with (dotted line) and without (solid line) clinical involvement of heart. (B) Patients with high (>50%) (•) and with low (<50%) (○) EVR₂₄. (C) Patients stratified according to high (>50%) (•) and low (<50%) (○) EVR₂₄ and presence (dotted line) or absence (solid line) of clinical involvement of heart.