The Reality of Type 2 Diabetes Prevention
The Reality of Type 2 Diabetes Prevention
The major prevention trials used either "lifestyle modification" or pharmacotherapy to significantly reduce the incidence of diabetes. Lifestyle modification was neither simple nor straightforward. In the Diabetes Prevention Program (DPP), the primary goal was a reduction of at least 7% of initial body weight and an enhanced level of physical activity in overweight or obese people at high risk of diabetes. In a post hoc study analysis of the DPP, weight loss was by far the most important contributor to prevention. Reduction of calories from fat and increased physical activity predicted and helped sustain weight loss. In the follow-up study of the DPP, termed the DPP Outcomes Study (DPPOS), the contribution of exercise to prevention went unmentioned, implying that this intervention was of marginal value in preventing diabetes. The Finnish Diabetes Prevention Study (DPS) not only strived for weight loss and increased physical activity in a similar population but also sought many changes in diet composition, such as a reduction in saturated fat and an increase in fiber intake. When all the variables were analyzed simultaneously "the only significant association was between weight reduction and diabetes risk". Although dietary composition and physical activity were important, "their effect was mediated through resulting weight reduction". Other randomized trials also show that weight loss is the key, if not critical, intervention to reduce diabetes risk. Thus, weight loss appears to take center stage when it comes to lifestyle modification delaying the onset of diabetes.
The Chinese Da Qing Diabetes Prevention study also showed great benefit in preventing diabetes through lifestyle modification, although the interventions were of questionable value. In that study, weight loss was minimal (~2 lbs) in the combined diet-exercise arm after the initial 6-year intervention; weight loss was actually greater in the control arm after 20 years of follow-up. Individuals randomized to the exercise-only arm showed a 46% reduction in the incidence of diabetes after 6 years. At baseline, however, the exercise group was already doing 42% more exercise than control subjects (casting some doubt on the randomization process); after the 6-year follow-up, their exercise level did not improve significantly from their baseline level. In the 20-year follow-up report, no information related to the exercise-only group was reported.
Despite the impressive results of the Chinese study (43% relative risk reduction in the lifestyle group after 20-years follow-up), it is impossible to translate the intervention used in this study since neither a complete description of the lifestyle changes that occurred nor the identification of the active component(s) are known. This point was made by the investigators in their discussion of the results.
In summary, it appears that weight loss is the key factor that reduces diabetes risk, and thus all efforts to translate the prevention trials to a community setting have focused on weight reduction. While exercise may be important, aside from the Chinese study results, which are inconclusive, we have no randomized controlled studies showing the benefit of exercise alone on diabetes prevention or what type of exercise or for how long (i.e., over months or years) is key. Ali et al. recently summarized all the lifestyle interventions initiated in real-world settings that have been modeled on the DPP. They identified 28 studies but only 4 were carried out for 12 months or longer, and none of those reported the incidence of diabetes.
In addition to the four longer-term studies identified by Ali et al., we performed a broad literature search using PubMed and Embase to identify randomized, controlled trials demonstrating the effectiveness of delivering behavioral weight loss interventions to overweight or obese adults in a community setting that had at least 1 year of follow-up.Table 2 shows the degree of weight loss seen in the intervention group over that seen in the control group in all the community studies we identified and in the two major prevention trials. There are two important points to be gleaned from these studies. One is that where data were collected beyond 1 year, weight regain begins, commencing about that time. Of note, however, in the studies by Katula et al., subjects regained virtually no weight in year 2, which is surprising given the fact that no other weight loss study using lifestyle modification was nearly as successful. The results from Katula et al. may, however, have been an artifact due to the fact the year 2 weight assessment was done on those returning for that examination (n = 127 or a 16% drop-out rate in the intervention arm; 11% drop-out rate in control subjects), whereas the baseline assessment was made on the entire starting cohort (n = 151).
Weight regain is very common in weight loss studies that use a behavior intervention. Thus, it is extremely difficult to maintain weight loss, even in studies where the intervention is still in full force and the enrollees are extremely well motivated. Weight regain is thought to occur because of the physiologic drive to return to one's previous weight.
The other key finding shown in Table 2 is that the degree of weight loss in the first year of the vast majority of the community programs was less than that achieved in the prevention trials at the equivalent period. Overall, weight loss in most community settings at year 1 was about the same as that achieved in the later years of the major trials. The most notable exception again was the studies by Katula et al., in which weight loss at year 2 was the closest to what was achieved in the study it attempted to translate (i.e., DPP). Notwithstanding potential issues related to the measurement of weight loss over the 2-year follow-up, as mentioned above, the results of the Katula studies' intervention may have promise if the weight loss achieved at year 2 can be sustained and replicated.
There are other reports that are relevant to weight loss in community settings. Finland has initiated a large-scale nationwide diabetes prevention program modeled after the DPS. The interim report of that effort showed that after 1 year, participants on average lost very little weight (mean 1.3%); only 18% of the cohort lost ≥5% of their body weight compared with nearly 50% of participants in the original DPS. Thus, the translation study could not achieve the same degree of weight loss as in the prevention trial. Compared with those who maintained their weight, people losing 2.5–4.9% body weight had only an 18% risk reduction, compared with a 58% relative risk reduction in the DPS after 3 years of follow-up. Also, there are widely available commercial weight loss programs that could be considered community-based interventions. These reports show that some participants can lose 3–4% of their body weight in the first year of enrollment.
The inability of nearly all the community studies to come close to replicating the weight loss achieved in year 1 of the prevention trials and the steady weight regain thereafter is troublesome. Moreover, as the lifestyle intervention in all the community studies was provided free, subjects received a great deal of staff-directed encouragement throughout the study and were likely to be a more highly motivated group. Such important benefits are unlikely to be provided outside of a study environment.
There appear to be no data directly documenting the relationship between weight loss and diabetes prevention. Hammen et al. reported that every kilogram of weight loss in the DPP resulted in a 16% reduction in risk, but their results were based entirely on data from the initial 3 years of the study. As the mean weight loss diminishes with time, so would the relationship between absolute weight reduction and risk of diabetes. In the DPS, the relative risk reduction in the cumulative incidence of diabetes was 36% when only the 3-year postintervention period was analyzed, which narrowly made statistical significance (P < 0.04). In the first year of the 3-year postintervention period (which corresponds to DPS year 5 data in Table 2 ), weight reduction was 2.1% relative to control subjects. It is very possible that as weight regain occurred in DPS (and DPP too); the cumulative incidence data over the initial years mask a much lower difference between groups in the annual incidence of diabetes in subsequent years. In fact, at year 4 of the DPS, fasting plasma glucose levels had returned to baseline values and continued to increase thereafter.
Other data support the hypothesis that weight loss of at least 3–4% over an extended period of time is necessary to achieve appreciable diabetes prevention. At the start of the DPPOS, the weight of individuals in the lifestyle arm was about 3–4% less than control subjects, and the incidence of diabetes was no different thereafter between the lifestyle and control arms. It was not entirely clear, however, whether the absence of a protective effect was due to the smaller percentage of weight lost relative to the DPP or due to other factors identified in the follow-up report. Also in the DPP/DPPOS, the mean A1C level in the lifestyle arm was lowest when weight loss was at its maximum (i.e., year 1); A1C levels increased thereafter, seeming to correlate with the degree of weight regain.
In a computer simulation of the DPP that relied only on the baseline data of the DPP participants and the degree of weight loss, the incidence of diabetes mimicked the results of the DPP and DPPOS. The simulation assumed a mean weight loss of 7% in year 1, gradually decreasing to 4% weight loss after 3 years (as seen in the DPP), and held at that level for decades. After 10 years of follow-up in the simulation, the incidence of diabetes in the lifestyle group was virtually the same as that achieved in the DPP. If the 4% weight loss in the simulation was maintained for 30 years, the risk of diabetes was reduced by a relative 15%. These data confirm that the benefit of lifestyle modification was due mainly to weight loss and that a 4% weight reduction for life would be beneficial.
Finally, in a study to determine whether bariatric surgery could prevent the onset of type 2 diabetes, after 15 years of follow-up there was an 83% relative reduction in the incidence of diabetes, and that corresponded to a relative 17–26% loss in body weight over the entire follow-up period.
Although the information above on the relationship between weight loss and diabetes prevention is not definitive, it does suggest that the delay in developing diabetes is relatively proportional to weight loss. Since nearly all the real-world studies could only achieve a weight loss in year 1 that corresponded to years 3–4 in the prevention trials and weight regain after year 1 is assured, it is likely that very few cases of diabetes would be prevented in community programs, particularly over more than just a couple of years. Unfortunately, none of the community studies were designed to determine the extent to which diabetes is prevented; all report only surrogate end points. So, in fact, we will not know the clinical benefit of modest weight loss anytime soon, but all the indirect evidence points to a greatly diminished, or absent, long-term benefit.
Many studies show that glucose-lowering drugs can delay the onset of diabetes. With the exception of metformin, none of the drugs have been given for more than 3 years. After their discontinuation, the incidence of diabetes increases. In the DPP, metformin was given for 3 years and resulted in a 31% reduction in the incidence of diabetes. The delay to the onset of diabetes was estimated to be half that achieved in the lifestyle group. About 75% of the participants in the metformin arm took what was considered to be the prescribed dose, but in the follow-up study the rate of adherence dropped to 57%. Considering safety and cost, metformin appears to be the best drug to reduce the incidence of diabetes, but success requires adherence to long-term treatment, which appears problematic and even less likely among individuals not enrolled in a clinical trial. While some investigators suggest using a combination of glucose-lowering drugs, such regimens have not been tested. Newer drugs hold promise in their ability to promote weight loss and improve cardiovascular risk factors, but they are very expensive and have not been given to individuals with prediabetes for long periods of time.
Despite the above considerations, if drugs are to be routinely used in people with prediabetes, then essentially we have moved the diagnostic cut point for diabetes to a lower level. While that may be warranted, all the ramifications of such a change deserve considerable discussion.
Can We Prevent or Delay Diabetes Outside of a Clinical Trial?
Lifestyle Modification
The major prevention trials used either "lifestyle modification" or pharmacotherapy to significantly reduce the incidence of diabetes. Lifestyle modification was neither simple nor straightforward. In the Diabetes Prevention Program (DPP), the primary goal was a reduction of at least 7% of initial body weight and an enhanced level of physical activity in overweight or obese people at high risk of diabetes. In a post hoc study analysis of the DPP, weight loss was by far the most important contributor to prevention. Reduction of calories from fat and increased physical activity predicted and helped sustain weight loss. In the follow-up study of the DPP, termed the DPP Outcomes Study (DPPOS), the contribution of exercise to prevention went unmentioned, implying that this intervention was of marginal value in preventing diabetes. The Finnish Diabetes Prevention Study (DPS) not only strived for weight loss and increased physical activity in a similar population but also sought many changes in diet composition, such as a reduction in saturated fat and an increase in fiber intake. When all the variables were analyzed simultaneously "the only significant association was between weight reduction and diabetes risk". Although dietary composition and physical activity were important, "their effect was mediated through resulting weight reduction". Other randomized trials also show that weight loss is the key, if not critical, intervention to reduce diabetes risk. Thus, weight loss appears to take center stage when it comes to lifestyle modification delaying the onset of diabetes.
The Chinese Da Qing Diabetes Prevention study also showed great benefit in preventing diabetes through lifestyle modification, although the interventions were of questionable value. In that study, weight loss was minimal (~2 lbs) in the combined diet-exercise arm after the initial 6-year intervention; weight loss was actually greater in the control arm after 20 years of follow-up. Individuals randomized to the exercise-only arm showed a 46% reduction in the incidence of diabetes after 6 years. At baseline, however, the exercise group was already doing 42% more exercise than control subjects (casting some doubt on the randomization process); after the 6-year follow-up, their exercise level did not improve significantly from their baseline level. In the 20-year follow-up report, no information related to the exercise-only group was reported.
Despite the impressive results of the Chinese study (43% relative risk reduction in the lifestyle group after 20-years follow-up), it is impossible to translate the intervention used in this study since neither a complete description of the lifestyle changes that occurred nor the identification of the active component(s) are known. This point was made by the investigators in their discussion of the results.
In summary, it appears that weight loss is the key factor that reduces diabetes risk, and thus all efforts to translate the prevention trials to a community setting have focused on weight reduction. While exercise may be important, aside from the Chinese study results, which are inconclusive, we have no randomized controlled studies showing the benefit of exercise alone on diabetes prevention or what type of exercise or for how long (i.e., over months or years) is key. Ali et al. recently summarized all the lifestyle interventions initiated in real-world settings that have been modeled on the DPP. They identified 28 studies but only 4 were carried out for 12 months or longer, and none of those reported the incidence of diabetes.
In addition to the four longer-term studies identified by Ali et al., we performed a broad literature search using PubMed and Embase to identify randomized, controlled trials demonstrating the effectiveness of delivering behavioral weight loss interventions to overweight or obese adults in a community setting that had at least 1 year of follow-up.Table 2 shows the degree of weight loss seen in the intervention group over that seen in the control group in all the community studies we identified and in the two major prevention trials. There are two important points to be gleaned from these studies. One is that where data were collected beyond 1 year, weight regain begins, commencing about that time. Of note, however, in the studies by Katula et al., subjects regained virtually no weight in year 2, which is surprising given the fact that no other weight loss study using lifestyle modification was nearly as successful. The results from Katula et al. may, however, have been an artifact due to the fact the year 2 weight assessment was done on those returning for that examination (n = 127 or a 16% drop-out rate in the intervention arm; 11% drop-out rate in control subjects), whereas the baseline assessment was made on the entire starting cohort (n = 151).
Weight regain is very common in weight loss studies that use a behavior intervention. Thus, it is extremely difficult to maintain weight loss, even in studies where the intervention is still in full force and the enrollees are extremely well motivated. Weight regain is thought to occur because of the physiologic drive to return to one's previous weight.
The other key finding shown in Table 2 is that the degree of weight loss in the first year of the vast majority of the community programs was less than that achieved in the prevention trials at the equivalent period. Overall, weight loss in most community settings at year 1 was about the same as that achieved in the later years of the major trials. The most notable exception again was the studies by Katula et al., in which weight loss at year 2 was the closest to what was achieved in the study it attempted to translate (i.e., DPP). Notwithstanding potential issues related to the measurement of weight loss over the 2-year follow-up, as mentioned above, the results of the Katula studies' intervention may have promise if the weight loss achieved at year 2 can be sustained and replicated.
There are other reports that are relevant to weight loss in community settings. Finland has initiated a large-scale nationwide diabetes prevention program modeled after the DPS. The interim report of that effort showed that after 1 year, participants on average lost very little weight (mean 1.3%); only 18% of the cohort lost ≥5% of their body weight compared with nearly 50% of participants in the original DPS. Thus, the translation study could not achieve the same degree of weight loss as in the prevention trial. Compared with those who maintained their weight, people losing 2.5–4.9% body weight had only an 18% risk reduction, compared with a 58% relative risk reduction in the DPS after 3 years of follow-up. Also, there are widely available commercial weight loss programs that could be considered community-based interventions. These reports show that some participants can lose 3–4% of their body weight in the first year of enrollment.
The inability of nearly all the community studies to come close to replicating the weight loss achieved in year 1 of the prevention trials and the steady weight regain thereafter is troublesome. Moreover, as the lifestyle intervention in all the community studies was provided free, subjects received a great deal of staff-directed encouragement throughout the study and were likely to be a more highly motivated group. Such important benefits are unlikely to be provided outside of a study environment.
How Much Weight Loss Protects Against Diabetes?
There appear to be no data directly documenting the relationship between weight loss and diabetes prevention. Hammen et al. reported that every kilogram of weight loss in the DPP resulted in a 16% reduction in risk, but their results were based entirely on data from the initial 3 years of the study. As the mean weight loss diminishes with time, so would the relationship between absolute weight reduction and risk of diabetes. In the DPS, the relative risk reduction in the cumulative incidence of diabetes was 36% when only the 3-year postintervention period was analyzed, which narrowly made statistical significance (P < 0.04). In the first year of the 3-year postintervention period (which corresponds to DPS year 5 data in Table 2 ), weight reduction was 2.1% relative to control subjects. It is very possible that as weight regain occurred in DPS (and DPP too); the cumulative incidence data over the initial years mask a much lower difference between groups in the annual incidence of diabetes in subsequent years. In fact, at year 4 of the DPS, fasting plasma glucose levels had returned to baseline values and continued to increase thereafter.
Other data support the hypothesis that weight loss of at least 3–4% over an extended period of time is necessary to achieve appreciable diabetes prevention. At the start of the DPPOS, the weight of individuals in the lifestyle arm was about 3–4% less than control subjects, and the incidence of diabetes was no different thereafter between the lifestyle and control arms. It was not entirely clear, however, whether the absence of a protective effect was due to the smaller percentage of weight lost relative to the DPP or due to other factors identified in the follow-up report. Also in the DPP/DPPOS, the mean A1C level in the lifestyle arm was lowest when weight loss was at its maximum (i.e., year 1); A1C levels increased thereafter, seeming to correlate with the degree of weight regain.
In a computer simulation of the DPP that relied only on the baseline data of the DPP participants and the degree of weight loss, the incidence of diabetes mimicked the results of the DPP and DPPOS. The simulation assumed a mean weight loss of 7% in year 1, gradually decreasing to 4% weight loss after 3 years (as seen in the DPP), and held at that level for decades. After 10 years of follow-up in the simulation, the incidence of diabetes in the lifestyle group was virtually the same as that achieved in the DPP. If the 4% weight loss in the simulation was maintained for 30 years, the risk of diabetes was reduced by a relative 15%. These data confirm that the benefit of lifestyle modification was due mainly to weight loss and that a 4% weight reduction for life would be beneficial.
Finally, in a study to determine whether bariatric surgery could prevent the onset of type 2 diabetes, after 15 years of follow-up there was an 83% relative reduction in the incidence of diabetes, and that corresponded to a relative 17–26% loss in body weight over the entire follow-up period.
Although the information above on the relationship between weight loss and diabetes prevention is not definitive, it does suggest that the delay in developing diabetes is relatively proportional to weight loss. Since nearly all the real-world studies could only achieve a weight loss in year 1 that corresponded to years 3–4 in the prevention trials and weight regain after year 1 is assured, it is likely that very few cases of diabetes would be prevented in community programs, particularly over more than just a couple of years. Unfortunately, none of the community studies were designed to determine the extent to which diabetes is prevented; all report only surrogate end points. So, in fact, we will not know the clinical benefit of modest weight loss anytime soon, but all the indirect evidence points to a greatly diminished, or absent, long-term benefit.
Pharmacotherapy
Many studies show that glucose-lowering drugs can delay the onset of diabetes. With the exception of metformin, none of the drugs have been given for more than 3 years. After their discontinuation, the incidence of diabetes increases. In the DPP, metformin was given for 3 years and resulted in a 31% reduction in the incidence of diabetes. The delay to the onset of diabetes was estimated to be half that achieved in the lifestyle group. About 75% of the participants in the metformin arm took what was considered to be the prescribed dose, but in the follow-up study the rate of adherence dropped to 57%. Considering safety and cost, metformin appears to be the best drug to reduce the incidence of diabetes, but success requires adherence to long-term treatment, which appears problematic and even less likely among individuals not enrolled in a clinical trial. While some investigators suggest using a combination of glucose-lowering drugs, such regimens have not been tested. Newer drugs hold promise in their ability to promote weight loss and improve cardiovascular risk factors, but they are very expensive and have not been given to individuals with prediabetes for long periods of time.
Despite the above considerations, if drugs are to be routinely used in people with prediabetes, then essentially we have moved the diagnostic cut point for diabetes to a lower level. While that may be warranted, all the ramifications of such a change deserve considerable discussion.
Source...