Gum Disease and the Rest of the Body

The instrument that defines gum disease is a slender steel probe marked in millimeters, and in the National Health and Nutrition Examination Survey the examiners walked it around the American mouth with bureaucratic thoroughness: six sites on every tooth, all teeth except the third molars, pocket depth and attachment loss recorded site by site. It is slow, unglamorous surveillance, and it produced one of the more arresting prevalence figures in chronic disease.

In the survey cycles of 2009 through 2012, 46% of dentate United States adults aged 30 and older met the case definition for periodontitis, a figure the CDC-led analysis translated to 64.7 million people; 8.9% had the severe form. Read at the level of the probe, 3.8% of all periodontal sites in the country, or 10.6% of all teeth, had pockets of 4 mm or deeper, and 19.3% of sites had lost 3 mm or more of attachment. The disease climbed with age and ran higher in men, and the burden was not evenly carried: prevalence reached 63.5% among Hispanic adults and 59.1% among non-Hispanic Black adults, against 50.0% among non-Hispanic Asian Americans and 40.8% among non-Hispanic whites, and it varied two-fold between the top and the bottom of the socioeconomic ladder, whether the ladder was measured by poverty or by education.

Periodontitis is an inflammatory disorder of the structures that support the teeth, and its endpoint, when it runs unchecked, is edentulism: no teeth at all. The interesting question has always been whether any of this stays in the mouth. The epidemiologists who compiled the broadest accounting in this record opened their report with a plain statement of where the field has arrived, writing that periodontitis “has been independently associated with the chronic noncommunicable diseases that most frequently lead to death worldwide.” The evidence behind that sentence, and the randomized trials that have put its causal reading to the test, reward a walk through in order.

The ledger of the dead

That accounting, published in the Journal of Dental Research by M. Romandini of Complutense University in Madrid and colleagues, pooled 57 studies covering 48 cohorts and 5.71 million participants. People with periodontitis died of all causes at 1.46 times the rate of people without it (95% confidence interval, 1.15 to 1.85). Cause by cause the pattern held and sharpened: cardiovascular death, 1.47; cancer death, 1.38; death from coronary heart disease, 2.58; death from cerebrovascular disease, 3.11. Pneumonia was the exception, at 0.98, which is to say no association at all.

Edentulism ran heavier still: 1.66 for death from any cause, 2.98 for coronary death, 3.18 for cerebrovascular death, and, unlike periodontitis itself, a significant 1.72 for pneumonia. The authors kept their conclusion restrained, writing that periodontitis and its ultimate sequela “are associated with an increased risk of all-cause and cause-specific mortality.” The operative word is “associated,” and the rest of this review is about how much weight that word can bear.

A smaller, harder number

Meta-analyses of mortality invite a fair objection, which is confounding: the people who accumulate periodontitis tend also to accumulate the standard cardiovascular exposures, smoking and diabetes among them, along with the disadvantages of income and education that the prevalence tables display so clearly. Any of those could be the real driver of the mortality curves, with the gums along for the ride.

A disciplined attempt to address that objection is a review conducted for the United States Preventive Services Task Force by Linda L. Humphrey and colleagues at the Oregon Evidence-based Practice Center. It admitted only prospective cohorts that had measured the Framingham risk factors and followed adults free of known coronary disease. Seven cohorts of good or fair quality survived the screening. Across every definition of exposure the review considered, from gingivitis and bone loss to periodontitis and tooth loss, the summary relative risks for incident coronary disease landed in a narrow band: 1.24 at the low end (95% confidence interval, 1.01 to 1.51), 1.34 at the high (1.10 to 1.63), and similar across gender, outcome and study quality. The verdict was written with a care worth preserving: periodontal disease “is a risk factor or marker for CHD that is independent of traditional CHD risk factors, including socioeconomic status.” The hedge in the middle of that sentence, “or marker,” holds the whole question.

The gap between this band of 1.24 to 1.34 and the mortality meta-analysis’s coronary 2.58 is not so much a contradiction as a change of subject. One measured new coronary disease in people who started clean; the other measured coronary death across every kind of cohort the literature offered, different outcomes counted in different studies across different decades. The honest reading is that the association is consistently present and that its size depends heavily on what is being counted.

The blood carries the news

If diseased gums reach the arteries, something has to carry the message, and the candidate these studies test is inflammation. C-reactive protein, which the Amsterdam group that compiled these numbers describes as a risk predictor for cardiovascular diseases, turns out to track the state of the gums. Spiros Paraskevas and colleagues at the Academic Centre for Dentistry Amsterdam screened 448 studies and kept 18 that measured high-sensitivity CRP in people with periodontitis and no other systemic disorder. Ten cross-sectional studies pooled to a weighted mean difference of 1.56 mg/l between patients and controls (p<0.00001), with patients often running above 2.1 mg/l. Six treatment studies pooled to a reduction of 0.50 mg/L after periodontal therapy (95% confidence interval, 0.08 to 0.93; p=0.02).

Their grading ran in two registers, and both are worth keeping: “strong evidence” that CRP runs higher in periodontitis, “modest evidence” that treating the gums brings it down. An inflamed mouth, on this record, is not a sealed compartment; the bloodstream reads it.

Worse before better

The clearest single experiment on the mouth-artery connection in this record appeared in the New England Journal of Medicine in 2007. Maurizio S. Tonetti, Francesco D’Aiuto and colleagues randomized 120 patients with severe periodontitis to community-based periodontal care (59 patients) or intensive periodontal treatment (61), then measured flow-mediated dilation of the brachial artery, the vessel’s dilation during increased flow, along with inflammatory and endothelial markers, at 1, 7, 30, 60 and 180 days after treatment.

The first result went the wrong way. Twenty-four hours after intensive treatment, flow-mediated dilation was significantly lower than in the control group, an absolute difference of 1.4% (95% confidence interval, 0.5 to 2.3; P=0.002), while C-reactive protein, interleukin-6 and the endothelial activation markers soluble E-selectin and von Willebrand factor all rose. Disturbing a severely infected mouth inflames, for a time, the body that houses it. Then the curve turned: by day 60 the treated group’s arteries dilated better than the controls’ (absolute difference 0.9%; 0.1 to 1.7; P=0.02), and by day 180 the advantage had grown to 2.0% (1.2 to 2.8; P<0.001), with the degree of arterial improvement tracking the degree of periodontal improvement (r=0.29; P=0.003).

The investigators reported no serious adverse effects and observed no cardiovascular events in either group, and their conclusion carried both halves of the story: intensive treatment produced “acute, short-term systemic inflammation and endothelial dysfunction,” while six months later the oral benefits “were associated with improvement in endothelial function.” A caution belongs here. Flow-mediated dilation is a physiologic surrogate, a measure of how a vessel behaves rather than of who goes on to have heart attacks, and this trial was not designed to count events. The artery’s behavior is one thread in cardiovascular risk among several; a companion review on this site examines what actually moves blood pressure.

Two trials, two answers

Blood sugar is where the causal question has been put to its most direct test in this record, partly because chronic periodontitis, in the words of one trial report, is prevalent in patients with diabetes, and partly because glycated hemoglobin hands a trial a clean, continuous endpoint. The two major randomized answers do not agree.

The Diabetes and Periodontal Therapy Trial, published in JAMA by Steven P. Engebretson and colleagues, was the larger of the two: 514 people with type 2 diabetes, hemoglobin A1c between 7% and just under 9%, and untreated moderate to advanced chronic periodontitis, enrolled at five American academic centers and randomized to scaling and root planing plus chlorhexidine rinse, with supportive care at 3 and 6 months, or to no treatment for six months. The dentistry worked as dentistry: probing depths came down by an adjusted 0.28 mm against the controls, attachment loss by 0.25 mm, bleeding on probing by 13.1%, all significant. The metabolism did not move: A1c drifted up 0.17% among the treated and 0.11% among the controls, a between-group difference of −0.05% (95% confidence interval, −0.23% to 0.12%; P=.55). Enrollment was stopped early for futility, and the investigators wrote the null result plainly, concluding that nonsurgical periodontal therapy “did not improve glycemic control.”

Five years later, in The Lancet Diabetes & Endocrinology, Francesco D’Aiuto’s group at the UCL Eastman Dental Institute in London reported the opposite outcome from a longer and deeper intervention: 264 patients with type 2 diabetes, moderate-to-severe periodontitis and at least 15 teeth, randomized to intensive treatment, meaning whole-mouth subgingival scaling with surgical periodontal therapy where oral hygiene allowed and supportive care every 3 months, or to supragingival scaling and polishing on the same schedule. Baseline A1c averaged 8.1% in both groups. At 12 months the intensively treated group came out 0.6% lower after adjustment (95% confidence interval, 0.3 to 0.9; p<0.0001), with serious adverse events reported in similar numbers in the two groups, and the authors concluded that “routine oral health assessment and treatment of periodontitis could be important for effective management of type 2 diabetes.”

The two trials differ in nearly every dimension that might matter: 6 months against 12; nonsurgical scaling against an intensive protocol that included surgery; a control arm left untreated against one given above-the-gumline cleaning; five centers against one; and the larger trial found nothing while the longer, deeper one found something. Whether treating the gums lowers blood sugar therefore depends, for now, on which design one believes better isolates the question. It is an open seam in the evidence, not a settled fact in either direction. The record at a glance:

An actuarial table with teeth

Where trials cannot reach, the population registers keep score, and two Scandinavian cohorts kept it for decades. In Gothenburg, Sweden, the gerontological population studies examined 1,803 people at the age of 70, drawn from four birth cohorts born between 1901 and 1922, and followed them through the national Swedish health registers. The examinations caught a great dental transition in a single variable: 51% of the earliest cohort had no teeth at 70; by the 1922 cohort the figure had fallen to 16%. Seven-year mortality ran 14% in women and 28% in men, and highest of all, at 42% and 47%, in the edentulous men of the last two cohorts.

Tor Osterberg and colleagues at the Sahlgrenska Academy adjusted their models for the standard suspects, health factors, socioeconomic position and lifestyle, and the count of teeth survived as an independent, statistically significant predictor of 7-year mortality in both sexes. Pooled across the four cohorts, the hazard ratio was 0.96 per tooth (95% confidence interval, 0.94 to 0.98; P<0.001, adjusted for cohort), which the authors translated into the plainest sentence in this record: “each remaining tooth at age 70 decreased the 7-year mortality risk by 4%.” Over 18 years, in the first three cohorts, the association held in men (0.97 per tooth), with a slighter estimate in women (0.98).

Copenhagen ran the longer watch. Poul Holm-Pedersen and colleagues took 573 non-disabled 70-year-olds, all born in 1914, sorted them by dentition into four bands (no teeth, one to nine, ten to nineteen, twenty or more), and re-examined the survivors at 5-year intervals for two decades; 78 remained at the 20-year mark, which is what following 70-year-olds for 20 years does to a cohort. Reaching 70 with one to nine teeth carried an odds ratio for onset of disability of 3.02 at the 10-year follow-up (95% confidence interval, 1.26 to 7.24), an association that adjustment for health and education did not dislodge. The edentulous carried a higher adjusted risk of death across 21 years (hazard ratio 1.26; 1.03 to 1.55). The authors closed on the line this whole literature keeps circling back to, suggesting that tooth loss may be “an early indicator of accelerated aging.”

The long road to the brain

The newest extension of the ledger runs toward dementia. The Atherosclerosis Risk in Communities study classified 8,275 participants, average age 63, by full-mouth periodontal examination, or as edentulous, between 1996 and 1998, then adjudicated dementia through 2016, an average follow-up of 18.4 years. Nineteen percent developed dementia, 1,569 people, a rate of 11.8 cases per 1,000 person-years. Ryan T. Demmer and colleagues reported in Neurology that severe periodontal disease carried an adjusted hazard ratio of 1.22 for incident dementia (95% confidence interval, 1.01 to 1.47) against periodontally healthy participants, while complete tooth loss came in at 1.21 with an interval that crossed one (0.99 to 1.48).

The pattern grew stranger, and more honest, when milder outcomes were included. For the combined endpoint of mild cognitive impairment or dementia, assessed in a subgroup of 4,559, edentulism ran strongest of all at 1.90 (1.40 to 2.58) while severe periodontitis, at 1.15 (0.88 to 1.51), was not significant, and the associations were stronger in participants 62 or younger at the outset (P for interaction, 0.02). The authors graded their own result with the right adverb, calling periodontal disease “modestly associated” with incident MCI and dementia. In a literature given to enthusiasm, the modesty is the credible part.

What does the full ledger add up to? The associations are broad, consistent in direction, and in places not small: earlier death, more coronary disease, more disability and more dementia among people with diseased gums or missing teeth, across millions of participants and every adjustment the epidemiologists could muster. The causal chain, tested where it can be tested, is real but short. Periodontal therapy was followed by an average fall in C-reactive protein, on evidence graded modest, and by improved arterial function within months under trial conditions, and in one of the two major trials it lowered blood sugar. None of the trials in this record was designed to test the far end of the chain, the end at which treating gums would be shown to postpone heart attacks, dementia or death. Until someone does, the mouth’s condition is best read the way the Task Force reviewers read it, as a risk factor or a marker, either of which makes it worth knowing.

Meanwhile the probe goes on making its rounds, six sites to a tooth, millimeter by millimeter. It was designed to measure the space between tooth and gum. On this evidence it is also reading something about the years ahead, whether or not the mouth is where those years are decided.

Educational, not medical advice.