Changes in cancer incidence in teenagers and young adults (ages 13 to 24 years) in England 1979‒2003

Authors

  • Robert D. Alston BSc, PhD,

    Corresponding author
    1. Cancer Research UK Pediatric and Familial Cancer Research Group, Royal Manchester Children's Hospital, University of Manchester, Stancliffe, Manchester, United Kingdom
    • Cancer Research UK Pediatric and Familial Cancer Research Group, Royal Manchester Children's Hospital, University of Manchester, Stancliffe, Hospital Road, Manchester M27 4HA UK===

    Search for more papers by this author
    • Fax (011) 44 161 922 2508

  • Marco Geraci BSc, PhD,

    1. Cancer Research UK Pediatric and Familial Cancer Research Group, Royal Manchester Children's Hospital, University of Manchester, Stancliffe, Manchester, United Kingdom
    Search for more papers by this author
  • Tim O. B. Eden MBBS,

    1. Academic Unit of Pediatric and Adolescent Oncology, Christie Hospital National Health Service Trust, University of Manchester, Withington, Manchester, United Kingdom
    Search for more papers by this author
  • Anthony Moran MB, MPH,

    1. North West Cancer Intelligence Service, Christie Hospital, Withington, Manchester, United Kingdom
    Search for more papers by this author
  • Steve Rowan BSc,

    1. National Cancer Intelligence Centre, Office for National Statistics, London, United Kingdom
    Search for more papers by this author
  • Jillian M. Birch BSc, MSc, PhD

    1. Cancer Research UK Pediatric and Familial Cancer Research Group, Royal Manchester Children's Hospital, University of Manchester, Stancliffe, Manchester, United Kingdom
    Search for more papers by this author

  • Data used in this study were contributed by the 9 regional cancer registries in England. Census output is Crown copyrighted and is reproduced with the permission of the Controller of Her Majesty's Stationary Office and the Queen's Printer for Scotland.

Abstract

BACKGROUND.

Cancer for teenagers and young adults represents a major source of morbidity and mortality. Trends in cancer incidence can provide pointers concerning how changes in the environment and in personal behavior affect cancer risks.

METHODS.

Data on 39,129 neoplasms in individuals ages 13 to 24 years who were diagnosed in England from 1979 to 2003 were analyzed. Variability in incidence by time period and differences in the time trends by age group, sex, and geographic region were analyzed using generalized linear models.

RESULTS.

Incidence rates of leukemias, lymphomas, central nervous system, bone, and germ cell tumors; melanoma; and carcinomas of the thyroid, ovary, cervix, and colon/rectum increased over time (all P < .01); whereas the incidence of carcinomas of the stomach and bladder decreased (both P < .01). These changes were consistent by age, sex, and region for most neoplasms. Melanoma incidence stabilized in southern England by 1993 but continued to increase in northern England (P = .001). The increase in non-Hodgkin lymphoma was greater in individuals ages 20 to 24 year than in younger individuals, but the increase in Hodgkin lymphoma was confined to individuals ages 13 to 14 years.

CONCLUSIONS.

The changes in incidence rates may have been caused in part by environmental changes and in part by behavioral changes in young individuals. Some of these results can be used to inform public health campaigns, which can be constructed to encourage better lifestyle choices by young individuals. Cancer 2008. © 2008 American Cancer Society.

Documenting cancer incidence patterns over time is important in planning clinical service provision. Such changes in incidence at the population level also may provide pointers to environmental, etiologic factors, offering the potential to develop preventive measures.

When analyzing large and heterogeneous groups of diseases in a single study, it is important that methods of analysis take into account the current scientific knowledge regarding grouping similar patients appropriately. Such an approach is applied through use of a specialized teenage and young adult cancer diagnostic classification scheme.1, 2 The use of this hierarchical scheme enables patients to be grouped appropriately throughout the analyses.

In a previous report, some consideration was given to changes in cancer incidence over time in teenagers and young adults.1 Incidence by region and socioeconomic deprivation also has been studied.3 This report expands the time frame, age range, and histologic detail of those analyses and also looks at how the changes over time vary by sex, age group, and geographic area.

MATERIALS AND METHODS

Incidence data on all newly diagnosed cases of malignancies in England from 1979 to 2003, inclusive, were supplied by the National Cancer Intelligence Centre, Office for National Statistics, London (ONS). Data items included the region of residence at diagnosis, age at diagnosis, year of diagnosis, sex, and coded diagnosis. Cases diagnosed from 1979 to 1994 were allocated International Classification of Diseases (ICD)–Ninth Revision, disease codes4 and International Classification of Diseases for Oncology –First Edition (ICD-O1)5 morphology codes. Cases from 1995 to 2003 were allocated ICD–10th Revision disease codes6 and ICD-O–Second Edition (ICD-O2) morphology codes.7 The patients diagnosed from 1979 to 1989 were assigned to the areas covered by Regional Health Authorities (RHAs), and patients diagnosed from 1990 to 2003 were assigned to Government Office Regions (GORs), as shown in Figure 1. This was because of changes in the geography of administrative regions for health service delivery.

Figure 1.

Melanoma incidence per million person-years at risk among individuals ages 13 to 24 years in England (a) from 1979 through 1983 by Regional Health Authority and b) from 1999 through 2003 by Government Office Region. E.Ang indicates East Anglia.

Cancers in individuals ages 13 to 24 years were grouped according to a morphology-based diagnostic scheme.1 The age range from 13 to 24 years was chosen because the distribution of cancer types in individuals ages 13 and 14 years is closer to that of older teenagers and adults aged < 24 years8 than the distribution in younger children and infants. All cases of malignancies except carcinoma of the skin were included. Nonmalignant intracranial and intraspinal neoplasms also were included, but all nonmalignant neoplasms at other sites were excluded. Subgroups were selected for presentation in the tables if at least 2 component subgroups had an overall rate of at least 1 per 1 million person-years at risk (mpyr).

Annual population estimates for England by sex and single year of age were obtained from the Population Estimates Unit of ONS. The underlying population and the numbers of patients in each cancer group were tabulated by time period, age group, and sex. The time periods for most analyses were 1979 to 1983, 1984 to 1988, 1989 to 1993, 1994 to 1998, and 1999 to 2003; however, for the analysis of time trend by region, the second period was extended (1984-1989 instead of 1984-1988), and the third period was shortened (from 1989-1993 to 1990-1993), because all patients who were diagnosed before 1990 were assigned to RHA areas, and all patients who were diagnosed during and after 1990 were assigned to GORs. The age groups were ages 13 to 14 years, 15 to 19 years, and 20 to 24 years, which corresponded to the available population data by region. These populations and patient counts were used to calculate incidence rates. Rates were standardized to the European standard population using the direct method.9 By using these estimates, a Poisson regression model was fitted for each cancer group, and the significance of the variability in incidence by time period was assessed after taking account of the variability in incidence by sex and age group.10 A regression curve was fitted assuming exponential changes in incidence between the periods; then, further models were fitted to explore whether the slope of this curve differed by sex, age group, and, if appropriate, cancer subtype and whether the changes in incidence deviated from an exponential trend. Separate analyses examining whether the time trends varied by region were conducted using RHA as the geographic variable for diagnoses from 1979 to 1989 and GOR as the geographic variable from 1990 to 2003.

RESULTS

Overall

There were 39,129 neoplasms in the analysis, and 54% occurred in young men (Table 1). Only 10% of the neoplasms occurred in individuals ages 13 to 14 years, and 56% occurred in individuals ages 20 to 24 years. Overall, lymphoma was the most common neoplasm in individuals ages 13 to 24 years, followed by carcinomas, central nervous system (CNS) tumors, and germ cell tumors (GCTs).

Table 1. Numbers of Cancers in Individuals Ages 13 to 24 Years in England From 1979 to 2003 by Age Group, Cancer Group, and Sex
Cancer GroupAge Group, ySexAll
13-1415-1920-24MenWomen
  1. CNS indicates central nervous system.

Leukemia81118791608258917094298
Lymphoma73934274948516139539114
CNS tumors89221772820312627635889
Bone tumors479119971914359622397
Soft tissue sarcomas228822101911369332069
Germ cell tumors1581313375646395885227
Melanoma897572239107820073085
Carcinomas28915144572175446216375
Other specified neoplasms77159202208230438
Unspecified neoplasms1878141101136237
All378013,32522,02421,22717,90239,129

The overall incidence rate was 193.8 (mpyr) (Table 2). This increased by 1.52% per annum (P < .0001) with a 95% confidence interval of 1.38% to 1.66%. The change was similar for young men and young women (P = .37) but varied by age group, with an annual increase of 1.1% both in individuals ages 13 to 14 years and in individuals ages 15 to 19 years, but the rate was 1.9% in individuals ages 20 to 24 years (P < .0001). The changes in cancer rates over time also demonstrated marked variation between the 10 main histologic groups (P < .0001) (Tables 2-5).

Table 2. Cancer Incidence Rates (per Million Person-Years at Risk) for All Neoplasms and Hematopoietic Neoplasms in Individuals Ages 13 to 24 Years in England 1979-2003
Cancer GroupRate per Million Person-Years at Risk by PeriodAnnual Percentage Change (95% CI)P for Variability in Incidence Over Time
1979-19831984-19881989-19931994-19981999-20031979-2003TrendNonlinearityBy SexBy Age Group
  1. 95% CI indicates 95% confidence interval.

All neoplasms163.6183.3197.5208.8224.0193.81.5 (1.4-1.7)<.0001.05.37<.0001
Leukemia20.020.921.822.422.821.50.7 (0.2-1.1).002.93.13.33
 Acute lymphoid leukemia10.411.010.911.011.410.90.4 (−0.2, −1.0).18.91.83.43
 Acute myelogenous leukemia6.67.27.58.28.07.41.0 (0.3-1.8).005.76.05.70
 Chronic myeloid leukemia1.51.51.71.92.51.82.8 (1.3-4.4).0002.68.78.73
 Other specified and unspecified leukemia1.61.21.81.30.91.3−2.0 (−3.7, −0.3).02.02.04.92
Lymphoma39.543.947.446.050.045.11.1 (0.8-1.3)<.0001.03.73.01
 Non-Hodgkin lymphoma10.212.314.415.016.813.62.4 (1.8-2.9)<.0001.29.13.04
 Hodgkin lymphoma29.331.633.031.033.331.60.5 (0.1-0.8).006.10.72.0007
Table 3. Cancer Incidence Rates (per Million Person-Years at Risk) for All for All Intracranial and Intraspinal Tumors in Individuals Ages 13 to 24 Years in England 1979-2003
Cancer GroupRate per Million Person-Years at Risk by PeriodAnnual Percentage Change (95% CI)P for Variability in Incidence Over Time
1979-19831984-19881989-19931994-19981999-20031979-2003TrendNonlinearityBy SexBy Age Group
  1. 95% CI indicates 95% confidence interval; CNS, central nervous system; NOS, not otherwise specified; PNET, primitive neuroectodermal tumor.

All CNS tumors26.827.731.130.531.129.30.8 (0.4-1.2)<.0001.15.84.91
Astrocytomas7.58.49.910.811.69.52.3 (1.6-2.9)<.0001.81.05.15
 Pilocytic astrocytoma0.80.91.33.03.51.88.7 (7-10.4)<.0001.02.01.79
 Other low-grade astrocytoma0.40.30.50.60.90.55.2 (2.4-8.2).0003.62.47.65
 High-grade astrocytoma0.81.11.42.42.91.66.9 (5.2-8.6)<.0001.40.91.53
 Astrocytoma, NOS5.46.16.84.84.35.5−1.2 (−2.0, −0.4).005.0004.22.04
Other gliomas4.04.13.73.33.63.7−0.8 (−1.9, 0.2).10.63.23.03
 Oligodendroglioma0.70.80.90.81.20.91.7 (−0.4, 3.8).12.55.32.93
 Other specified gliomas0.30.30.50.81.10.67.8 (4.9-10.8)<.0001.93.66.06
 Glioma, NOS2.92.92.31.71.32.3−3.9 (−5.2, −2.6)<.0001.33.18.14
Ependymoma1.21.61.31.61.71.51.4 (−0.2, 3.0).09.39.15.11
Medulloblastoma/PNET1.71.81.52.12.21.91.3 (−0.1, 2.7).08.41.36.38
 Medulloblastoma1.10.80.71.00.80.9−0.8 (−2.9,1.3).44.16.38.44
 Noncerebellar PNET0.71.00.91.11.41.03.2 (1.2-5.3).002.70.83.76
Other specified CNS tumors9.99.210.811.19.710.10.3 (−0.3, 1.0).27.03.85.70
 Craniopharyngioma1.51.41.11.21.11.3−1.7 (−3.4, 0.1).06.88.85.30
 Pituitary tumors3.23.14.14.32.83.50.2 (−0.8, 1.3).65.0006.68.62
 Pineal tumors0.80.60.50.50.40.6−2.7 (−5.3, −0.1).04.86.49.78
 Choroid plexus tumors0.10.10.10.10.10.11.6 (−4.2, 7.8).59.72.67.25
 Meningioma1.50.81.51.61.31.30.5 (−1.2, 2.2).60.008.55.78
 Nerve sheath tumors of the brain2.22.52.82.12.32.40.1 (−1.1, 1.4).85.23.73.13
 Hemangioblastomas and gangliogliomas0.70.60.71.41.61.05.5 (3.4-7.6)<.0001.16.73.04
Unspecified CNS tumors2.52.84.01.62.22.6−1.0 (−2.2, 0.2).10<.0001.39.26
Table 4. Cancer Incidence Rate (per Million Person-Years at Risk) for Soft Tissue Sarcomas, Bone Tumors, Germ Cell Tumors, and Melanoma in Individuals Ages 13 to 24 Years in England 1979-2003
Cancer GroupRate per Million Person-Years at Risk by PeriodAnnual Percentage Change (95% CI)P for Variability in Incidence Over Time
1979-19831984-19881989-19931994-19981999-20031979-2003TrendNonlinearityBy SexBy Age Group
  • 95% CI indicates 95% confidence interval; NOS, not otherwise specified.

  • *

    P value by sex not applicable.

Bone tumors11.011.411.813.413.112.01.0 (0.5-1.6).0003.59.72.02
 Osteosarcoma6.16.15.66.65.96.10.1 (−0.8, 0.8).96.39.60.06
 Chondrosarcoma0.81.00.70.70.90.8−0.5 (−2.6, 1.7).67.26.99.04
 Ewing sarcoma3.23.43.95.35.44.13.1 (2.1-4.1)<.0001.27.84.33
  Of bone3.13.13.43.63.73.31.0 (0.0-2.1).06.97.41.96
  Extraskeletal0.00.20.41.51.30.614.4 (11-17.9)<.0001.0001.54.10
  Unspecified site0.00.10.00.20.40.111.6 (5.3-18.3).0001.45.71.47
 Other bone tumors0.81.01.70.80.91.00.1 (−1.8, 2.1).91.001.831.00
Soft tissue sarcomas (STS)8.610.911.29.910.910.30.7 (0.1-1.4).02.003.23.51
 Fibromatous neoplasms2.52.62.01.51.82.1−2.3 (−3.6, −0.9).001.24.88.76
 Rhabdomyosarcoma1.83.02.62.22.12.4−0.1 (−1.3, 1.2).89.002.002.87
 Other specified STS3.03.44.34.75.04.02.7 (1.7-3.7)<.0001.64.62.41
 STS, NOS1.32.02.31.51.91.81.0 (−0.5, 2.4).20.004.63.87
Germ cell tumors (GCT)18.922.725.529.833.625.72.9 (2.5-3.3)<.0001.7.01.95
 Testicular GCT29.938.541.149.756.342.33.0 (2.6-3.5)<.0001<.0001*.93
 Ovarian GCT4.13.94.85.46.24.82.3 (1.0-3.6).0004.008*.01
 CNS GCT0.60.41.31.31.30.95.3 (3.2-7.5)<.0001.009.11.37
 GCT of other sites1.10.91.11.00.91.0−0.6 (−2.5, 1.4).56.65.03.19
Melanoma8.913.216.118.920.315.13.8 (3.3-4.3)<.0001.0004.71.10
Table 5. Cancer Incidence Rates (per Million Person-Years at Risk) for Carcinomas in Individuals Ages 13 to 24 Years in England 1979-2003
Cancer GroupRate per Million Person-Years at Risk by PeriodAnnual Percentage Change (95% CI)P for Variability in Incidence Over Time
1979-19831984-19881989-19931994-19981999-20031979-2003TrendNonlinearityBy SexBy Age Group
  • 95% CI indicates 95% confidence interval; GU, genitourinary; GI, gastrointestinal; NEC, not elsewhere classified.

  • *

    P value by sex not applicable.

All carcinomas27.329.328.134.639.331.31.8 (1.4-2.2)<.0001.0002.0001.47
Head and neck           
 Thyroid4.55.65.57.98.46.23.2 (2.4-4.1)<.0001.16.79.43
 Nasopharyngeal carcinoma1.41.11.21.11.41.2−0.2 (−1.9, 1.6)0.84.39.63.01
 Other sites in lip or oral cavity1.21.51.62.12.61.83.8 (2.3-5.3)<.0001.92.15.27
 Other and ill-defined head and neck0.50.10.30.20.20.3−2.7 (−6.4, 1.1).16.12.14.24
 Trachea, bronchus, and lung0.90.70.90.91.30.92.4 (0.3-4.6).02.25.13.15
Breast2.22.72.92.72.62.60.7 (−0.5, 1.9).26.37.46.26
GU tract           
 Kidney0.90.71.01.10.80.90.5 (−1.6, 2.6).66.14.79.21
 Bladder1.21.51.21.00.61.1−3.0 (−4.8, −1.1).002.04.17.73
 Ovary4.66.04.98.810.06.74.0 (2.9-5.2)<.0001<.0001*.15
 Cervix8.79.69.210.712.110.01.6 (0.7-2.5).0004.005*.15
 Other and ill-defined GU tract0.80.60.60.50.40.6−3.3 (−5.9, −0.1).01.87.13.41
GI tract           
 Colon and rectum2.62.82.53.65.53.33.9 (2.8-5.1)<.0001.001.28.68
 Stomach0.70.70.50.30.50.6−3.5 (−6.1, −0.8).01.14.55.40
 Liver and intrahepatic bile duct0.80.81.10.91.31.02.1 (0.1-4.2).04.40.331.00
Pancreas0.30.30.40.20.20.3−1.7 (−5.5, 2.2).38.48.98.99
 Other and ill-defined GI tract0.30.40.20.40.40.31.4 (−2.4, 9.0).43.14.10.47
Other and ill-defined sites NEC2.62.11.41.82.12.0−1.3 (−2.6, 0.1).07.006.19.07

Leukemias

Leukemia incidence increased steadily at a rate of 0.7% per annum (P = .002) (Table 2). There were statistically significant increases in chronic myeloid leukemia (CML) (P = .0002) and acute myelogenous leukemia (AML) (P = .005) with a nonsignificant increase in acute lymphoid leukemia (ALL) (P = .18), but the difference in annual percentage change between ALL, CML, and AML was not statistically significant (P = .15). The incidence of AML rose more rapidly in young women (1.8% per annum) than in young men (0.3% per annum) (P = .05). The incidence in the other and unspecified leukemia group fell, which was attributed entirely to a decrease in the proportion of unspecified leukemias.

Lymphoma

The rate of non-Hodgkin lymphoma (NHL) increased at 2.4% per year (P < .0001) (Table 2), and a greater increase was observed among individuals ages 20 to 24 years (3.1%) than among individuals ages 15 to 19 years (1.8%) or ages 13 to 14 years (1.5%) (P = .04). The overall Hodgkin lymphoma (HL) incidence rate increased more slowly than the rate for NHL (P < .0001) at 0.5%, but that increase was significant (P = .006). In contrast to the patterns for NHL, the increase in HL was confined to individuals ages 13 to 14 years, for whom the rise was 2.8% per annum (P < .0001 for age rate differences).

Central Nervous System

The overall CNS tumor incidence rate rose at 0.8% per annum (Table 3), a trend that differed by morphologic group (P < .0001) but not by age or sex. Astrocytoma incidence increased by 2.3% (P < .0001) (Table 3), and other glioma incidence decreased slightly at 0.8%; however, taken together, the incidence rate of gliomas and astrocytomas increased by 1.4% per annum. The proportion of astrocytomas that were classified as astrocytoma not otherwise specified (NOS) dropped from 73% during 1979 through 1983 down to 37% during 1999 through 2003, and a similar change was observed in the other glioma group. Ependymoma and medulloblastoma/primitive neuroectodermal tumor exhibited nonsignificant increases. Craniopharygioma and pineal tumors both had marginally significant decreases in incidence. The increase in the gangliogliomas and hemangioblastomas group was caused entirely by an increase in gangliogliomas, whereas hemangioblastomas had a stable incidence. The incidence rate of unspecified CNS tumors fluctuated with a peak in the third period.

Bone Tumors

The incidence rate for bone tumors had an increase of 1% per annum (Table 4). Ewing sarcoma increased at 3.1% per annum, but this varied by site (P < .0001), and there were much larger increases in extraskeletal and unspecified site Ewing sarcoma than in Ewing sarcoma located in bone. The incidence of osteosarcoma and chondrosarcoma was stable throughout the period.

Soft Tissue Sarcomas

The overall rate of soft tissue sarcomas incidence increased slightly (0.7% per annum) (Table 4), but there was a peak in the period 1989 through 1993. This trend varied between the subgroups (P < .0001). Fibromatous neoplasms decreased in incidence, rhabdomyosarcomas had stable incidence, and the incidence of other specified soft tissue sarcomas increased. Kaposi sarcoma increased from 0.05 mpyr in 1979 through 1983 up to 0.6 mpyr in 1989-1993 before decreasing to 0.3 mpyr in 1999-2003 (data not shown).

Germ Cell Tumors

The incidence rate of GCTs increased at 2.9% per year (Table 4), but this varied by site (P = .0002). There were comparable increases (P = .37) in ovarian GCTs (2.3%) and testicular GCTs (3%). Other site GCTs decreased slightly at 0.6% per annum. The incidence of CNS GCTs demonstrated a step change at around 1991 but was constant during the periods 1979 through 1988 and 1989 through 2003.

Melanoma

The rate of melanoma nearly doubled between the period 1979 through 1983 and the period 1989 through 1993 (Table 4), but more recent increases were much less dramatic.

Carcinomas

Carcinoma incidence increased at a rate of 1.8% per year (Table 5) (P < .0001), with particularly large increases observed in the most recent periods (P = .0002), but this varied between sites (P < .0001). The overall incidence increased much more rapidly among young women than among young men (P = .0005). This difference was because of rapid increases in thyroid carcinoma (78% of those carcinomas occurred in young women) and in ovarian and cervical carcinomas. For each of the 17 sites considered, the trends were generally similar across the age range and by sex. The sole exception was for nasopharyngeal carcinoma, which demonstrated a decrease in incidence of 4.1% for individuals ages 13 and 14 years and an increase of 3.2% for individuals ages 20 to 24 years.

Cervical carcinomas had a consistent increase during the periods 1994 through 1998 and 1999 through 2003, but not before. Ovarian carcinoma incidence was greater during the period using ICD-O2 coding (1995-2003) than earlier. Bladder carcinoma incidence decreased. There was a recent, large increase in colorectal carcinoma (3.9% per annum overall; P < .0001) but a decrease in stomach carcinoma of 3.5% per annum (P = .01). Carcinomas in other and unspecified sites decreased at 1.3% per annum (P = .07).

Other Specified and Unspecified Tumors Not Elsewhere Classified

The overall rate of other specified tumors did not change during the study period. The overall rate of tumors that were unspecified increased from 0.5 per mpyr in 1979 through 1983 to 1.9 per mpyr 1989 through 1993, but then declined to 0.9 per mpyr. Overall, only 0.6% of neoplasms were unclassified, and this rate never rose above 1% in any period.

Time Trends by Region

The time trends were consistent between regions for all of the main cancer groups (data not shown) except melanoma. The melanoma rate during 1979 through 1983 was greatest in the southernmost regions (P = .0001) (Fig. 1a), and the increase in the period 1984 through 1989 was consistent across the regions (P = .85) (Tables 6, 7). Between the periods 1990 through 1993 and 1999 through 2003, the changes in incidence differed strongly by region (P = .001), with more rapid increases in the northernmost regions until, in the latest period (Fig. 1b), the highest rates were reported in the north of England.

Table 6. Melanoma Incidence Rates by Time Period and Regional Health Authorities (1979-1989) in England
PeriodMelanoma Incidence Rate: Regional Health Authorities
NorthernNWMerseyYorksTrentW.MidE.AngThamesWessexOxfordSW
NWNESESW
  1. W.Mid indicates West Midlands, E.Ang, East Anglia.

1979-19837.57.28.05.66.88.38.210.56.07.213.819.09.312.2
1984-198913.010.29.711.610.411.711.114.68.912.714.222.115.219.9
Table 7. Melanoma Incidence Rates by Time Period and Government Office Regions (1990-2003) in England
PeriodMelanoma Incidence Rate: Government Office Regions
North EastNorth WestYorks and HumberEast MidlandsWest MidlandsEast of EnglandLondonSouth EastSouth West
1990-199310.018.419.113.914.115.414.520.722.0
1994-199815.421.922.818.316.119.612.518.424.5
1999-200331.026.225.516.216.817.011.421.025.1

DISCUSSION

To our knowledge, this is the first report to examine in detail temporal trends in national cancer incidence among individuals ages 13 to 24 years in England and makes use of data collected from all 9 regional cancer registries over a 25-year period. With a population of 49 million, England is the most populous country in the world, with a national high-quality cancer registration system.11 Cancer registration data from the middle 1970s onward generally are complete and of high quality.12 The interregional variability in incidence was small3 for all groups except melanoma, and the time trends observed were consistent between registries. Only ovarian carcinomas had a step change in incidence that matched the change in coding system from ICD-O1 to ICD-O2. The results for groups other than ovarian carcinoma cannot be explained reasonably by the change in coding regime or by differences in registration practices either between registries or over time.

In the current study, the overall increases over time for respective age ranges were similar to those reported among individuals ages 15 to 19 years in other European countries13 and among individuals ages 15 to 24 years in the US,14 but there were important differences by cancer type. Neither of those studies covered the entire teenage and young adult age range,13 used a childhood cancer classification scheme,14 or used a mixture of a morphology-based childhood cancer classification scheme and a site-based adult cancer classification scheme.

Melanoma is 1 of the few major cancers in which the principle cause in adults is both known and potentially avoidable. Previous studies have noted striking differences in melanoma incidence across all ages in England and Wales over time15 by socioeconomic deprivation and by region.9 The patterns by deprivation and region for individuals ages 13 to 24 years follow patterns similar to those reported at all ages.3 The overall increase of 3.8% per annum reported in the current study was much greater than the 1.2% reported among individuals ages 20 to 24 years in the US over the period from 1975 to 2000.14 The incidence of invasive melanoma in adults aged < 35 years was stable in New South Wales, Australia during the period 1983 to 1996, but the incidence at older ages continued to increase.16 In Canada, incidence among individuals ages 20 to 39 years increased from 1969 to 1984 but declined during the period 1985 to 1993,17 and again continued to increase among older individuals. The current results for southern England are similar to those for Canada, albeit with a later end to the increase. Because exposure to ultraviolet radiation is the greatest risk factor for melanoma, the changing geographic distribution of melanoma may be related to differing levels of exposure in the area of residence and changes in holiday destination patterns for UK residents.18 With the increase in air travel and the opening of regional airports, holidays to Spain and the Mediterranean have become more popular throughout the UK.

It remains to be determined whether the changes observed in the geographic pattern of melanoma among younger individuals in different regions of England has also occurred at older ages. This merits further study.

The use of a morphology-based classification scheme allows comparisons of morphologically similar neoplasms, even when these are at different sites and may form only a small proportion of all neoplasms at certain sites. The close similarity of the rate of increase of ovarian and testicular GCTs observed here among teenagers and young adults also was observed in an all-ages analysis from southeast England.19 The similarity of the overall age distribution of cases,19 incidence trends, and biologic traits implies that these tumors may have common etiologies and, possibly, common risk factors, although the rarity of ovarian GCT makes detailed studies difficult. Ovarian GCT rates reportedly increased in the US (1973-2002) among individuals ages 15 to 19 years, but not among older or younger individuals,20 whereas our data, with considerably more incident cases, indicated a consistent increase in ovarian GCTs throughout the age range from 13 to 24 years. However, the patterns observed in CNS GCT incidence, with an apparent step change around 1989 but with stable rates before and after that date, are suggestive of a change in diagnostic methods, including the increased use of better radiologically guided biopsies.

The increases in the incidence rates of CNS tumors in England up to 2003 generally were similar to the increases noted among individuals ages 15 to 19 years during 1978 through 1997 in a variety of regions in Europe21 and among individuals ages 15 to 29 years during 1975 through 2000 in the Surveillance, Epidemiology, and End Results (SEER) registries in the US.14 The increases also were similar to those reported in children ages birth to 14 years in North West England.22 In contrast to CNS GCTs, the changes in incidence rates over time, both overall (P = .15) and for most other types, were more gradual. The more detailed delineation of CNS tumors with less explicit diagnoses, such as glioma NOS, both from exactly specified neoplasms (such as oligodendroglioma) and from unspecified CNS neoplasms, in the classification scheme that we used led to improvements in diagnostic specificity and enabled us to recognize these entities more clearly than was possible with the schemes used in the other studies. Given the decrease in the proportion of cases with less explicit diagnoses, it is extremely difficult to estimate changes in the underlying rates of specific tumors, particularly the very rare types.

In general, the changes in rates for other solid tumors reflected patterns observed among older individuals in England15 and among individuals ages 15 to 24 years in the US.14 The incidence of soft tissue sarcomas and bone sarcomas appeared to be stable for both children22 and teenagers/young adults, as reported here. This, together with the small-scale, spatially uniform incidence and the lack of association with ecologic risk factors in children23, 24 and in teenagers/young adults,3 implies that the etiologic mechanisms for most sarcomas in this age group may be related to intrinsic factors that affect growth, development, and genetic susceptibility.25 The slight peak observed in Kaposi sarcoma may be attributed to the human immunodeficiency virus/acquired immunodeficiency syndrome epidemic, and its subsequent decline may have been caused by preventive measures and antiretroviral treatments. The rise in reported nonbone Ewing sarcoma incidence is most likely the result of improvements in molecular pathology and diagnostic classification of small round cell tumors rather than actual underlying incidence.

The rise in thyroid carcinoma incidence also was observed among older individuals in England12 and Scotland.26 It was considerably greater than the rate of increase observed in the US14 among individuals ages 15 to 24 years, although the overall incidence in England was lower than for the US.14 There was no evidence of an effect of the Chernobyl nuclear incident in 1986 on either the overall rate or for interregional differences in the rates. The increase in cervical carcinoma rates, which is linked to the human papillomavirus (HPV),27 may be related to general increases in sexually transmitted infections in this age group.28 The interval between initial infection and the development of carcinoma in these young women presumably is quite short. Recent evidence29 suggests that the rise in squamous cell carcinomas of the head and neck may be related to HPV.

Ovarian carcinoma incidence had a step change, with an increase in incidence around 1995, which is when ICD-O2 coding was introduced. Approximately 55% of tumors that were classified as ovarian carcinomas and coded with the ICD-O2 scheme were cystadenomas of borderline malignancy, which were excluded in the ICO-O1-based analyses. When the analysis was restricted to those entities that were considered malignant in both coding regimes, the incidence was stable (data not shown). The pattern was different from that observed with ovarian GCTs, demonstrating the advantage of using a scheme that uses both morphology and site information for the classification of cancers in this age range.

All ages incidence of leukemia has been stable in England and Wales since the 1970s.15 The overall rate of increase observed here in England among individuals ages 13 to 24 years was similar to that from the SEER registries among individuals ages 15 to 29 years from 1979 to 2000.14 The estimation of annual percentage change of specific subtypes of leukemia is complicated by the rapid decline in the proportion of cases registered as leukemia NOS. The incidence of lymphomas and overall trends by age were similar to those reported in the SEER study.14 However, the findings in the current study that NHL was increasing more rapidly among individuals ages 20 to 24 years than among younger individuals and HL was increasing more rapidly among individuals ages 13 to 14 years than among older individuals appear to be new observations.

Cancer represents a major source of morbidity and mortality among individuals ages 13 to 24 years and is responsible for 10.8% of deaths in this age range in the UK. Trends in the incidence of disease, when based on data from reliable population-based registries, are important in informing both professionals and the public about which diseases are likely to pose major public health problems in the future and, for those diseases with a known etiology, allow action at both a societal and individual level to reduce risks. In particular, cervical carcinoma and melanoma, which potentially are avoidable, relatively easy to diagnose in their early stages, and have a much better prognosis when diagnosed early, are both rising rapidly in incidence in this age range.

Ancillary