The source population comprised patients who received care from the Marshfield Clinic, a large multispecialty group practice with 54 locations in rural central, western, and northern Wisconsin. The clinic developed and maintains an electronic vaccination registry (www.recin.org) for all immunizations administered by Marshfield Clinic providers, in addition to those given by many local public health agencies and immunization providers.

Two cohorts comprising 685 subjects were enrolled over a 1-year period. Cohort 1 comprised 113 young adult subjects who participated in a 12-year longitudinal study at the Marshfield Clinic examining immunogenicity and adverse events following the second dose of MMR vaccine, hereafter called the “longitudinal study” [23, 24]. To achieve adequate sample size, cohort 2 was recruited. Cohort 2 comprised 572 young adults identified using Marshfield's vaccination registry who had 2 documented doses of MMR vaccine but did not participate in the longitudinal study. Invitation letters were mailed to both cohorts, and follow-up phone calls were made.

Although only 25 (22.1%) cohort 1 subjects had low or negative mumps titers at any point during the longitudinal study, all 93 cohort 1 subjects with at least 1 low or negative titer to any of the 3 antigens during the longitudinal study (defined as <121 mIU/mL for measles [25], ≤10 mIU/mL for mumps [23], or ≤10 mIU/mL for rubella [26]) and all cohort 2 subjects were offered a third dose of MMR vaccine (M-M-R II, Merck & Co) in this study. We combined cohorts 1 and 2 for analysis purposes as there were no statistically significant differences between the 2 cohorts in terms of sex, race/ethnicity, age, geometric mean titers (GMTs) at baseline or 1 year post-MMR3, or percentage with negative or low baseline mumps titers (Supplementary Table 1). Serum was collected from these study subjects immediately before, and 1 month and 1 year after MMR3.

At each visit, subjects were questioned concerning mumps disease, exposures, vaccinations, and other health events. MMR vaccine was administered during the initial visit according to a standard protocol. Adverse events were evaluated and will be reported elsewhere. Informed consent was obtained by all subjects. The study was approved by the institutional review boards of the Marshfield Clinic Research Foundation and the Centers for Disease Control and Prevention (CDC).

The analysis of data for all 3 antigens was taken into consideration when determining sample size. To detect a decrease in the proportion of subjects who had low or negative titers from the last draw of the longitudinal study compared with 1 year post-MMR3, we based our sample size on a decrease from approximately 20% with low or negative titers at the last draw of the longitudinal study in 2006–2007 to 10% 1 year post-MMR3 for mumps, 5% to 1% for measles, and 50% to 30% for rubella, with 90% power and 95% confidence intervals. The target sample size of 375 was increased to 685 to account for the fact that more than one-third of the 312 subjects from the longitudinal study were ineligible to receive a third dose based on high titers for all 3 antigens throughout the longitudinal study and 53% attrition during the longitudinal study [23, 25, 26].

Subjects were excluded if they had a history of measles, mumps, or rubella disease, lived in the same household with anyone who had these diseases during the subject's lifetime, previously received a third MMR vaccine dose, received any vaccinations within 30 days of enrollment, had any contraindications to MMR vaccination, or had any condition likely to impair immune response, as specified in the ACIP recommendations [27].

Although there is no established correlate of immunity for mumps, neutralizing antibody is likely essential for protection against mumps and is considered the gold standard for mumps serology. Thus, plaque reduction neutralization (PRN) assay was used to determine virus neutralizing antibody titer in sera as described previously [28, 29]. Heat-inactivated sera were serially diluted 2-fold from 1:4 to 1:128 and mixed with an equal volume of the Jeryl Lynn vaccine virus diluted to contain approximately 80 plaque-forming units (PFU), resulting in a final serum dilution range of 1:8 to 1:256. Virus control wells were incubated with the virus preparation and an equal volume of minimal essential media (MEM) containing 5% fetal bovine serum (FBS). Reference serum “Lot 3” was included in each assay run. Following a 1-hour incubation period, half of each of the virus/serum mixtures (containing approximately 40 PFU of virus) was transferred to each of 2 wells in 24-well plates containing Vero cell monolayers and overlaid with 2% carboxymethylcellulose (Sigma) in MEM supplemented with 10% FBS. After 5 days of incubation at 37°C, wells were stained with neutral red (Sigma), and plaques were counted the following day. The mean plaque number was determined for duplicate wells at each serum dilution. The neutralizing antibody titer was calculated as the serum dilution capable of reducing the mean number of virus plaques by ≥50% compared to the mean number of plaques in virus control wells using the Kärber formula [30]. Sera not reaching a 50% endpoint were retested in assays using a higher dilution series.

No established PRN mumps titer correlates with mumps immunity [31]. Therefore, the cutoffs used in this study for seronegative, low-positive, and high-positive were chosen based on a previous study [29]. Mumps virus neutralizing antibody titers <1:8 (limit of assay detection) were considered seronegative. For analysis purposes, we considered titers between 1:8 and <1:16 (ie, within a single dilution factor of the limit of detection) to be low-seropositive and titers ≥1:16 to be high-seropositive.

Data were pooled across assay runs that met the following 2 validity criteria: (1) the mean plaque number in the negative serum control wells was between 20 and 60 (a range validated in the laboratory to not influence measured neutralizing antibody titers), and (2) the neutralizing antibody titer for reference mumps serum Lot 3 was required to be within 2 standard deviations (SD) of its GMT based on historic data. Assays not meeting these poolability requirements were retested. Serum samples from individual subjects were tested in the same assay run. Other than each subject's unique identifier code and serum collection dates, laboratories were blinded to study information.

Mantel-Haenszel χ² and Fisher exact tests were run to assess categorical variables. Wilcoxon rank-sum tests were used for continuous variables. Variables that were considered potential risk factors for low or negative mumps titers (defined as <16 mIU/mL in this study) included age at first MMR dose, time since second MMR dose (we used <15 years vs ≥15 years prior based on the average age of subjects at enrollment minus the age when the second dose was recommended), sex, race/ethnicity, military member, post–secondary school attendance, number of household members, current illnesses, current medications, and (for post-MMR3 serum samples) the binary variable of whether the subject had low or negative baseline mumps titers. In multivariate logistic regression, a forward stepwise selection approach that used P values <.4 for inclusion and <.05 for retention identified whether any factors were independently associated with negative or low mumps antibody titer levels at baseline, 1 month and 1 year post-MMR3. GMTs were calculated from base 2 log-transformed data. Statistical significance was assigned for P values <.05. Data were analyzed with SAS software, version 9.3. Reverse cumulative distribution curves were created by SigmaPlot 12 (Systat Software, Inc) and were used to compare the shift in the curves from baseline, 1 month, and 1 year.

From the longitudinal study, we successfully contacted 194 of 200 persons attempted. Of those, 113 (58%) were enrolled, 45 (23%) refused, and 36 (19%) were ineligible (15 due to prior receipt of MMR3 and 21 for other reasons). To achieve adequate sample size, we attempted to contact an additional 1795 persons and successfully reached 1379 (77%). Of those, 572 (41%) were enrolled, 664 (48%) refused, and 143 (10%) were ineligible (4 due to prior MMR3 receipt and 139 for other reasons).

Baseline serum samples were obtained from 678 of 685 subjects enrolled from the combined group of longitudinal study participants and new recruits; 656 (95.8%) received MMR3 and completed at least 1 follow-up draw. There were 655 (99.5%) subjects who completed the 1-month draw and 612 (93.3%) who completed the 1-year draw. We excluded 20 (2.9%) subjects who were not given MMR3, because the group was too small to be considered a comparison group. An additional 2 (0.3%) were excluded because they only had baseline data. We analyzed data from 656 subjects (Figure 1); 290 (44.2%) were male and 644 (98.2%) were self-declared non-Hispanic white. Subjects ranged in age from 18 to 28 years, (mean, 20.8 [SD, 2.1] years). Figure 1.

Flow chart for enrollment, analysis, and vaccination of subjects with a third dose of measles-mumps-rubella (MMR) vaccine.

At baseline, 5 (0.8%) subjects were seronegative, 38 (5.8%) were low-seropositive, and 613 (93.4%) were high-seropositive (Figure 2A). Of the 613 subjects with high-seropositive baseline titers, 612 had sera drawn at 1 month and 572 had sera drawn at 1 year; all remained high-seropositive. Of the 5 subjects who were seronegative at baseline, 1 became low-seropositive and 4 became high-seropositive 1 month after MMR3. One year post-MMR3, the low-seropositive subject returned to seronegative status, while 2 high-seropositive subjects at 1 month became low-seropositive. Of the 38 low-seropositive subjects at baseline, 29 were high-seropositive 1 month post-MMR3, 9 were low-seropositive, and none were negative. One year post-MMR3, 35 of 38 had serum samples drawn, of whom 19 were high-seropositive, 14 were low-seropositive, and 2 were seronegative. Figure 2.

A, Flow chart of mumps antibody titer levels at baseline, 1 month, and 1 year. B, Percentage of subjects who had negative, low-seropositive, and high-seropositive mumps antibody titer levels at baseline, and 1 month and 1 year following a third dose of measles-mumps-rubella (MMR) vaccine.

GMTs were statistically different between baseline and 1 month post-MMR3 (104.1 vs 159.2; P < .0001), as well as between baseline and 1 year post-MMR3 (104.1 vs 125.9; P < .01). However, as shown in the reverse cumulative distribution curves (Figure 3), the shift in mumps titers from baseline to 1 month to 1 year was minimal. The shape and distribution of the curves at all 3 time points was nearly identical when using untransformed mumps titers and transformed titers on a logarithmic scale. Figure 3.

Reverse cumulative distribution curve using log-transformed titers by percent of subjects who had neutralizing mumps antibody titers at baseline and 1 month and 1 year following a third dose of measles-mumps-rubella (MMR3) vaccine.

Forty of 655 (6.1%) subjects had 4-fold rises from baseline to 1 month post-MMR3, of whom 4 had negative baseline titers, 11 had low baseline titers, and 25 had positive baseline titers. Thirteen of 612 (2.1%) subjects had 4-fold rises from baseline to 1 year post-MMR3, of whom 2 had negative baseline titers, 3 had low baseline titers, and 8 had positive baseline titers. There were no subjects with a 4-fold rise in titer from 1 month to 1 year.

By χ² analysis, sex, race/ethnicity, military member, post–secondary school attendance, number of household members, current illnesses, and current medications were not associated with negative or low-seropositive mumps titers at baseline, 1 month, or 1 year post-MMR3 (Table 1). Table 1.

Risk Factors for Negative or Low Mumps Titers at Baseline and 1 Month and 1 Year After Receiving the Third Measles-Mumps-Rubella Vaccine Dose

Significant risk factors for negative or low baseline mumps titers by χ² analysis were age at first MMR dose (odds ratio [OR] = 2.58; 95% confidence interval [CI], 1.08–6.13; P = .03) and time since second MMR dose (OR = 0.38; 95% CI, 0.16–0.92; P = .03). Of the 50 (7.6%) subjects who received their first dose at age 12 to <15 months, 7 (14%) had negative or low baseline mumps titers, compared with 36 of 606 (5.9%) subjects who were vaccinated with their first dose at age ≥15 months. Of the 189 (28.8%) subjects who received their second dose <15 years prior, 6 (3.2%) had negative or low titers, whereas, of the 467 (71.2%) subjects who received their second dose ≥15 years prior, 37 (7.9%) had negative or low titers.

By χ² analysis, a significant risk factor for negative or low mumps titers 1 month post-MMR3 was whether a subject had low or negative baseline mumps titers (OR, 384.0; 95% CI, 22.0–6692.9; P < .0001). Significant risk factors for negative or low mumps titers 1 year post-MMR3 were whether a subject had low or negative baseline mumps titers (OR, 1038.5; 95% CI, 60.7–17 773.6; P < .0001) and the time since the second MMR dose (OR, 0.38; 95% CI, 0.16–0.92, P = .02). When mumps titer levels were assessed as a continuous variable, baseline MMR3 titers were significantly associated with individual log-transformed mumps titer levels at 1 month and 1 year. Subjects with lower baseline titers were more likely to have lower titers at 1 month and 1 year, whereas subjects with higher baseline titers were more likely to have higher titers at 1 month and 1 year (R² = 0.81–0.87; P < .0001; Figure 4). Figure 4.

A, Comparison of individual mumps titer levels at baseline and 1 month following a third dose of measles-mumps-rubella (MMR) vaccine. Circles represent individual titer levels. The dark solid line represents the linear regression of the best-fit of the comparison. The light shading around the line represents the 95% confidence interval. The dotted lines represent the 95% confidence limit. B, Comparison of individual mumps titer levels at baseline and 1 year following a third dose of MMR vaccine. Circles represent individual titer levels. The dark solid line represents the linear regression of the best fit of the comparison. The light shading around the line represents the 95% confidence interval. The dotted lines represent the 95% confidence limit.